Replace zipfile with unzip

6 files changed, 107 insertions, 39 deletions

diff --git a/lib/document-to-html.js b/lib/document-to-html.js
index fe3b566..64672ee 100644
--- a/lib/document-to-html.js
+++ b/lib/document-to-html.js
@@ -97,9 +97,9 @@ function DocumentConverter(options) {
             });
         },
         "image": function(element, html, callback) {
-            element.read().then(function(imageBuffer) {
+            element.read("base64").then(function(imageBuffer) {
                 // TODO: set alt attribute to wp:inline/wp:docPr/@descr
-                var src = "data:image/png;base64," + imageBuffer.toString("base64");
+                var src = "data:image/png;base64," + imageBuffer;
                 var attributes = {src: src};
                 if (element.altText) {
                     attributes.alt = element.altText;
diff --git a/lib/document-xml-reader.js b/lib/document-xml-reader.js
index ea15204..04eec21 100644
--- a/lib/document-xml-reader.js
+++ b/lib/document-xml-reader.js
@@ -116,9 +116,7 @@ function DocumentXmlReader(relationships, docxFile) {
                 .first("a:blip")
                 .attributes["r:embed"];
             var imagePath = relationships[relationshipId].target;
-            var readImage = function() {
-                return docxFile.read(path.join("word", imagePath));
-            };
+            var readImage = docxFile.read.bind(docxFile, path.join("word", imagePath));
             var altText = element.first("wp:docPr").attributes.descr;
             
             var image = documents.Image(readImage, altText)
diff --git a/lib/docx-reader.js b/lib/docx-reader.js
index ecaa10c..9cb912f 100644
--- a/lib/docx-reader.js
+++ b/lib/docx-reader.js
@@ -2,8 +2,9 @@ exports.read = read;
 
 var q = require("q");
 var _ = require("underscore");
-var ZipFile = require("zipfile").ZipFile;
+var unzip = require("unzip");
 var path = require("path");
+var fs = require("fs");
 
 var xmlreader = require("./xmlreader");
 var documents = require("./documents");
@@ -13,39 +14,95 @@ var Result = require("./results").Result;
 
 function read(docxFile) {
     if (_.isString(docxFile)) {
-        var zipFile = new ZipFile(docxFile);
-        var exists = function(path) {
-            return zipFile.names.indexOf(path) !== -1;
-        }
-        docxFile = {
-            read: function(path, encoding) {
-                if (exists(path)) {
-                    return q.ninvoke(zipFile, "readFile", path)
-                        .then(function(value) {
-                            if (encoding) {
-                                return value.toString(encoding);
-                            } else {
-                                return value;
-                            }
-                        });
-                } else {
-                    return q.reject(new Error("No file in zip: " + path));
-                }
-            },
-            exists: exists
-        };
+        var zipPath = docxFile;
+        docxFile = openZip(zipPath);
     }
+    return q.when(docxFile, function(docxFile) {
     
-    return q.all([
-        readXmlFile(docxFile, "word/_rels/document.xml.rels"),
-        readXmlFile(docxFile, "word/document.xml")
-    ]).spread(function(relationshipsXml, documentXml) {
-        var relationships = relationshipsXml ? readRelationships(relationshipsXml) : {};
-        var reader = new DocumentXmlReader(relationships, docxFile);
-        return reader.convertXmlToDocument(documentXml);
+        return q.all([
+            readXmlFile(docxFile, "word/_rels/document.xml.rels"),
+            readXmlFile(docxFile, "word/document.xml")
+        ]).spread(function(relationshipsXml, documentXml) {
+            var relationships = relationshipsXml ? readRelationships(relationshipsXml) : {};
+            var reader = new DocumentXmlReader(relationships, docxFile);
+            return reader.convertXmlToDocument(documentXml);
+        });
     });
 }
 
+function openZip(zipPath) {
+    var deferred = q.defer();
+    var names = [];
+    
+    fs.createReadStream(zipPath)
+        .pipe(unzip.Parse())
+        .on("entry", function(entry) {
+            names.push(entry.path);
+            entry.autodrain();
+        })
+        .on("close", function() {
+            deferred.resolve(zipFile);
+        })
+        .on("error", function(err) {
+            deferred.reject(err);
+        });
+    
+    var exists = function(path) {
+        return names.indexOf(path) !== -1;
+    }
+    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;
+            
+            if (exists(path)) {
+                return q.ninvoke(zipFile, "readFile", path)
+                    .then(function(value) {
+                        if (encoding) {
+                            return value.toString(encoding);
+                        } else {
+                            return value;
+                        }
+                    });
+            } else {
+                return q.reject(new Error("No file in zip: " + path));
+            }
+        },
+        exists: exists
+    };
+    return deferred.promise;
+}
+
 function readXmlFile(docxFile, path) {
     if (docxFile.exists(path)) {
         return docxFile.read(path, "utf-8")
diff --git a/package.json b/package.json
index 3b815f2..cb9a44b 100644
--- a/package.json
+++ b/package.json
@@ -13,7 +13,7 @@
         "sax": "~0.5.2",
         "underscore": "~1.4.4",
         "lop": "~0.2.6",
-        "zipfile": "~0.3.4",
+        "unzip": "~0.1.7",
         "optimist": "~0.3.5",
         "connect": "~2.7.3",
         "dustjs-linkedin": "~1.2.1",
diff --git a/tests/document-to-html.tests.js b/tests/document-to-html.tests.js
index 04ac143..5ed2224 100644
--- a/tests/document-to-html.tests.js
+++ b/tests/document-to-html.tests.js
@@ -158,8 +158,8 @@ describe('DocumentConverter', function() {
     test('images are written with data URIs', function() {
         var imageBuffer = new Buffer("Not an image at all!");
         var image = new documents.Image(
-            function() {
-                return q.when(imageBuffer);
+            function(encoding) {
+                return q.when(imageBuffer.toString(encoding));
             }
         );
         var converter = new DocumentConverter();
diff --git a/tests/testing.js b/tests/testing.js
index af8ec94..6fef626 100644
--- a/tests/testing.js
+++ b/tests/testing.js
@@ -1,6 +1,7 @@
 var path = require("path");
 var fs = require("fs");
 var q = require("q");
+var _ = require("underscore");
 
 exports.test = test;
 exports.testData = testData;
@@ -22,8 +23,20 @@ function testData(testDataPath) {
 }
 
 function createFakeDocxFile(files) {
-    function read(path) {
-        return q.when(files[path]);
+    function read(path, encoding) {
+        return q.when(files[path], function(buffer) {
+            if (_.isString(buffer)) {
+                buffer = new Buffer(buffer);
+            }
+            
+            if (!Buffer.isBuffer(buffer)) {
+                return q.reject(new Error("file was not a buffer"));
+            } else if (encoding) {
+                return q.when(buffer.toString(encoding));
+            } else {
+                return q.when(buffer);
+            }
+        });
     }
     
     function exists(path) {