KEEL-66: archive extractor improvements - extract only text files

package.json

@@ -17,9 +17,11 @@
     "is-gzip": "^1.0.0",
     "is-tar": "^1.0.0",
     "is-zip": "^1.0.0",
+    "istextorbinary": "^1.0.2",
     "jade": "~1.9.2",
     "log4js": "^0.6.26",
     "mime": "^1.3.4",
+    "mkdirp": "~0.5.0",
     "morgan": "~1.5.3",
     "multer": "0.1.6",
     "node-schedule": "^0.2.9",
@@ -30,8 +32,7 @@
     "request": "^2.57.0",
     "rimraf": "^2.4.0",
     "tar": "^2.1.1",
-    "unzip": "^0.1.11",
-    "mkdirp": "~0.5.0"
+    "unzip": "^0.1.11"
   },
   "devDependencies": {
     "assert": "^1.3.0",

wrapper/archiveExtractor.js

@@ -13,6 +13,7 @@ var isTar = require('is-tar');
 var zlib = require('zlib');
 var tar = require('tar');
 var recursiveReaddir = require('recursive-readdir');
+var isTextOrBinary = require('istextorbinary');
 
 function ArchiveExtractor() {
     var self = this;
@@ -64,7 +65,7 @@ function ArchiveExtractor() {
                         return callback(err, session);
                     }
 
-                    if (isFile) {
+                    if (isFile && isTextOrBinary.isTextSync(entry.path)) {
                         var uniqid = randomstring.generate(10);
                         var filePath = path.join(directory, uniqid);
                         //logger.debug('Filepath: ' + filePath);

wrapper_configs/global.js

@@ -217,7 +217,29 @@ config.availableWappers = {
         id: 'uzip',
         port: 3007,
         class: 'archiveExtractor',
-        requestConf: simpleCommandRequest
+        requestConf: {
+            requestBodyTemplate: {
+                //isAsync: null
+            },
+            requestBodyParamsMappings: {
+                //isAsync: {
+                //    usageType: config.paramUsageTypes.META,
+                //    filter: function(value){
+                //        return value == 1;
+                //    },
+                //    required: true,
+                //    allowEmpty: false,
+                //    validator: function(value, request){ return true; }
+                //}
+            },
+            requestFiles: {
+                content: null
+            },
+            staticParams: {
+                sessionMaxLifetime: 600,
+                isAsync: 0
+            }
+        }
     },
     TOKENIZER : {
         title: 'Sõnestaja',