Apache Tika bridge. Text extraction, metadata extraction, mimetype detection and language detection.


Provides text extraction, metadata extraction, mime-type detection, text-encoding detection and language detection. All via a native Java bridge with the Apache Tika content-analysis toolkit. Bundles Tika 1.10.

Depends on node-java, which itself requires the JDK and Python 2 (not 3) to compile.

Requires JDK 7. Run node version to check the version that node-java is using. If the wrong version is reported even if you installed JDK 1.7, make sure JAVA_HOME is set to the correct path then delete node_modules/java and rerun npm install.

var tika = require('tika');
var options = {
    // Hint the content-type. This is optional but would help Tika choose a parser in some cases. 
    contentType: 'application/pdf'
tika.text('test/data/file.pdf', options, function(errtext) {

We can even extract directly from the Web. If the server returns a content-type header, it will be passed to Tika as a hint.

tika.text('http://www.ohchr.org/EN/UDHR/Documents/UDHR_Translations/eng.pdf', function(errtext) {
    // ... 

Or extract text using OCR (requires Tesseract).

tika.text('test/data/ocr/simple.jpg', {
    ocrLanguage: 'eng'
}, function(errtext) {
    // ... 

All methods that accept a uri parameter accept relative or absolute file paths and http:, https: or ftp: URLs.

The available options are the following.

  • contentType to provide a hint to Tika on which parser to use.
  • outputEncoding to specify the text output encoding. Defaults to UTF-8.
  • password to set a password to be used for encrypted files.
  • maxLength to specify a max number of character to extract.
  • ocrLanguage to set the language used by Tesseract. This option is required to enable OCR.
  • ocrPath to set the path to the Tesseract binaries.
  • ocrMaxFileSize to set maximum file size in bytes to submit to OCR.
  • ocrMinFileSize to set minimum file size in bytes to submit to OCR.
  • ocrPageSegmentationMode to set the Tesseract page segmentation mode.
  • ocrTimeout to set the maximum time in seconds to wait for the Tesseract process to terminate.
  • pdfAverageCharTolerance see PDFTextStripper.setAverageCharTolerance(float).
  • pdfEnableAutoSpace to set whether the parser should estimate where spaces should be inserted between words (true by default).
  • pdfExtractAcroFormContent to set whether content should be extracted from AcroForms at the end of the document (true by default).
  • pdfExtractAnnotationText to set whether to extract text from annotations (true by default).
  • pdfExtractInlineImages to set whether to extract inline embedded OBX images.
  • pdfExtractUniqueInlineImagesOnly as multiple pages within a PDF file might refer to the same underlying image.
  • pdfSortByPosition to set whether to sort text tokens by their x/y position before extracting text.
  • pdfSpacingTolerance see PDFTextStripper.setSpacingTolerance(float).
  • pdfSuppressDuplicateOverlappingText to set whether the parse should try to remove duplicated text over the same region.
  • pdfUseNonSequentialParser to set whether to use PDFBox's non-sequential parser.

Extract both text and metadata from a file.

tika.extract('test/data/file.pdf', function(errtextmeta) {
    console.log(text); // Logs 'Just some text'. 
    console.log(meta.producer[0]); // Logs 'LibreOffice 4.1'. 

Extract text from a file.

tika.text('test/data/file.pdf', function(errtext) {

Get an XHTML representation of the text extracted from a file.

tika.xhtml('test/data/file.pdf', function(errxhtml) {

Extract metadata from a file. Returns an object with names as keys and arrays as values.

tika.meta('test/data/file.pdf', function(errmeta) {
    console.log(meta.producer[0]); // Logs 'LibreOffice 4.1'. 

Detect the content-type (MIME type) of a file.

tika.type('test/data/file.pdf', function(errcontentType) {
    console.log(contentType); // Logs 'application/pdf'. 

Detect the character set (text encoding) of a file.

tika.charset('test/data/file.txt', function(errcharset) {
    console.log(charset); // Logs 'ISO-8859-1'. 

Detect the content-type and character set of a file.

The character set will be appended to the mime-type if available.

tika.typeAndCharset('test/data/file.txt', function(errtypeAndCharset) {
    console.log(typeAndCharset); // Logs 'text/plain; charset=ISO-8859-1'. 

Detect the language a given string is written in.

tika.language('This is just some text in English.', function(errlanguagereasonablyCertain) {
    console.log(language); // Logs 'en'. 
    console.log(reasonablyCertain); // Logs true or false. 

Developed by Matthew Caruana Galizia at the ICIJ.

Please feel free to submit an issue or pull request. Don't forget to add your name to the CONTRIBUTORS file.

Copyright (c) 2013 The Center for Public Integrity®. See LICENSE.

Apache Tika JAR distributed under the Apache License, Version 2.0.