fixes #60, removing unnecessary call to preprocess content

dbashford · Nov 23, 2015 · c28ec1f · c28ec1f
1 parent 936b890
commit c28ec1f
Show file tree

Hide file tree

Showing 7 changed files with 4 additions and 31 deletions.
diff --git a/lib/extractors/docx.js b/lib/extractors/docx.js
@@ -27,7 +27,7 @@ var _calculateExtractedText = function( inText ) {
     text += localText + "\n";
   });
 
-  return util.replaceTextChars( text );
+  return text;
 };
 
 var extractText = function( filePath, options, cb ) {

diff --git a/lib/extractors/odt.js b/lib/extractors/odt.js
@@ -41,10 +41,8 @@ var extractText = function( filePath, options, cb ) {
           for (var i = 0; i < nodes.length; i++) {
             nodeTexts.push($(nodes[i]).text());
           }
-          var contentWithBreaks = nodeTexts.join("\n");
-          contentWithBreaks = util.replaceTextChars(contentWithBreaks);
 
-          cb( null, contentWithBreaks );
+          cb( null, nodeTexts.join("\n") );
         });
       }
     });

diff --git a/lib/util.js b/lib/util.js
@@ -1,11 +1,5 @@
 var exec = require('child_process').exec;
 
-var SINGLE_QUOTES = /[\u2018|\u2019]/g
-  , DOUBLE_QUOTES = /[\u201C|\u201D]/g
-  , MULTI_SPACES = /[^\S\r\n]{2,}/g
-  , NON_ASCII_CHARS = /[^\x00-\x7F\x80-\xFF]/g
-  ;
-
 var yauzlError = function( err, cb ) {
   var msg = err.message;
   if ( msg === "end of central directory record signature not found" ) {
@@ -37,13 +31,6 @@ var unzipCheck = function(type, cb) {
   );
 };
 
-var replaceTextChars = function(text) {
-  return text.trim()
-    .replace( SINGLE_QUOTES, "'" )
-    .replace( DOUBLE_QUOTES, '"' )
-    .replace( MULTI_SPACES, ' ' );
-};
-
 var getTextFromZipFile = function( zipfile, entry, cb ) {
   zipfile.openReadStream( entry, function( err, readStream ) {
     if ( err ) {
@@ -74,7 +61,6 @@ var getTextFromZipFile = function( zipfile, entry, cb ) {
 module.exports = {
   createExecOptions: createExecOptions,
   unzipCheck: unzipCheck,
-  replaceTextChars: replaceTextChars,
   getTextFromZipFile: getTextFromZipFile,
   yauzlError: yauzlError
 };
diff --git a/test/cli_test.js b/test/cli_test.js
@@ -8,7 +8,6 @@ describe("cli", function(){
   it("will extract text", function(done) {
     exec( cliPath + " " + testFilePath,
       function( error, stdout, stderr ) {
-        console.log(stdout)
         expect(stdout).to.eql(".foo {color:red}\n");
         done();
       }

diff --git a/test/extract_test.js b/test/extract_test.js
@@ -269,7 +269,7 @@ describe('textract', function() {
       fromFileWithPath(filePath, function( error, text ) {
         expect(error).to.be.null;
         expect(text).to.be.a('string');
-        expect(text).to.eql( "this is a test document that won't be extracted properly." );
+        expect(text).to.eql( "this is a test document that won't be extracted properly. " );
         done();
       });
     });

diff --git a/test/url_test.js b/test/url_test.js
@@ -1,6 +1,7 @@
 var path = require("path");
 
 describe("fromUrl tests", function() {
+  this.timeout(3000);
 
   var test = function(ext, name, _text) {
     it('will ' + ext + ' files', function(done) {

diff --git a/test/util_test.js b/test/util_test.js