Apache Tika¶

What it is¶

The Apache Tika toolkit is an open-source library that detects and extracts metadata and text from over a thousand different file types (such as PPT, XLS, and PDF). It provides a single, unified interface for parsing complex document formats, making it a foundational tool for search engines and content analysis systems.

What problem it solves¶

Processing diverse file formats requires specialized libraries for each type (e.g., PDFBox for PDFs, POI for Office docs). Tika simplifies this by abstracting the complexity behind a single API. It solves the problem of "dark data" by allowing automated systems to "see" inside binary files to extract searchable text and structured metadata.

Where it fits in the stack¶

Category: Service / Data Processing. It sits in the data ingestion and extraction layer, acting as a pre-processor that converts unstructured binary documents into structured text that can be indexed by search engines (like Solr or Elasticsearch) or processed by LLMs.

Typical use cases¶

Search Indexing: Extracting text from uploaded PDFs and Word docs for a searchable knowledge base.
Content Analysis: Identifying the language and metadata (author, creation date) of large file archives.
RAG Pre-processing: Converting diverse document formats into plain text for use in Retrieval-Augmented Generation pipelines (often feeding into Ollama or LiteLLM).
Attachment Processing: Automatically extracting text from email attachments for automated routing or classification in n8n.

Strengths¶

Universal Parser: Supports an incredible range of formats out of the box.
Metadata Rich: Extracts not just text, but deeply embedded metadata.
Language Detection: Built-in ability to identify the language of the extracted text.
OCR Integration: Can automatically trigger Tesseract OCR for images or "image-only" PDFs.
Modular Architecture (v3.0+): Highly customizable parser configurations and reduced footprint for specialized deployments.

Limitations¶

Java Based: The core library and server run on the JVM, which can be memory-intensive. Requires Java 11+ as of version 3.0.
Formatting Loss: Focuses on text extraction; original visual formatting and layout are typically lost.
Complex Setup: Configuring specialized parsers (like OCR or deep object recognition) can require significant effort.

When to use it¶

When you need to extract text and metadata from a wide variety of file formats (PPT, XLS, PDF, etc.).
For search engine indexing where you need consistent output across different document types.
For content analysis, translation, and language detection tasks.
When building automated document processing pipelines.

When not to use it¶

If you only need to process a single, specific format and a lightweight, specialized library exists (e.g., just plain text).
In extremely memory-constrained environments, as the Java-based Tika server can be resource-intensive.
If you require the highest possible OCR accuracy and speed, specialized OCR engines may be more appropriate (though Tika integrates with Tesseract).

Getting started¶

Docker installation (Tika 3.0 Baseline)¶

The easiest way to run Tika Server is via Docker:

docker run -d -p 9998:9998 --name tika apache/tika:3.0.0.0

Tika will be available at http://localhost:9998.

Java Application (CLI)¶

You can also use the Tika application jar for local processing without a server:

# Download the latest tika-app jar (v3.0.0)
curl -O https://archive.apache.org/dist/tika/3.0.0/tika-app-3.0.0.jar

# Run it against a file
java -jar tika-app-3.0.0.jar --text document.pdf

Hello World¶

Ensure the Tika Docker container is running: docker ps | grep tika
Create a test file: echo "Hello Tika World" > test.txt
Send it to Tika's REST API: curl -T test.txt http://localhost:9998/tika
Tika should return the extracted text: Hello Tika World

CLI examples¶

You can use the Tika Server's REST API via curl or use the tika-app JAR for command-line processing.

# Extract text from a local PDF file using Tika Server
curl -T document.pdf http://localhost:9998/tika

# Extract metadata from a local document in JSON format
curl -H "Accept: application/json" -T document.docx http://localhost:9998/meta

# Detect the MIME type of a file
curl -T image.png http://localhost:9998/detect/stream

# List all available parsers (using tika-app)
java -jar tika-app.jar --list-parsers

API examples¶

You can interact with the Tika Server using any HTTP client.

Python (using requests)¶

import requests

url = "http://localhost:9998/tika"
headers = {"Accept": "text/plain"}

with open("document.pdf", "rb") as f:
    response = requests.put(url, data=f, headers=headers)

print(response.text)

Recursive Metadata (curl)¶

The /rmeta endpoint is powerful for getting metadata and text from both the container and all embedded documents (e.g., images inside a Word doc).

curl -T document.docx http://localhost:9998/rmeta/text --header "Accept: application/json"

n8n Integration: Automated PDF-to-Markdown¶

Tika is frequently used as a core node in n8n workflows to convert incoming attachments into a format suitable for LLM processing or archival.

Workflow Pattern: Email Attachment to Markdown¶

IMAP Email Trigger: Watch for new emails with PDF attachments.
HTTP Request Node (Tika):
- Method: PUT
- URL: http://tika:9998/tika
- Body: Attachment binary data.
- Headers: Accept: text/plain.
Code Node (Formatting): Clean up the raw text and wrap it in Markdown headers or metadata blocks.
Vector Store / Storage: Send the Markdown text to Supabase or save it as a file in Nextcloud.

Example Code Node (JS)¶

// Clean up Tika output for Markdown
const rawText = $node["HTTP Request"].data["text"];
const fileName = $node["IMAP Email"].data["attachment"]["name"];

const markdown = `
# Document: ${fileName}
## Extracted Content
${rawText.replace(/\n{3,}/g, '\n\n')}
`;

return [{ json: { markdown } }];

Unstructured.io — a modern alternative for document processing in AI pipelines
Paperless-ngx — uses Tika/OCR for document archival and searching
Ollama — for processing Tika-extracted text with local LLMs
Whisper — for complementary audio/video transcription
ChangeDetection.io — can trigger Tika when specific document links change
Elasticsearch — often used to store and search the text extracted by Tika
n8n — for building automated workflows that trigger Tika on new file uploads
Pandoc — for converting between document formats (Tika extracts, Pandoc converts)
Tesseract OCR — the engine Tika uses for optical character recognition

Apache Tika¶

What it is¶

What problem it solves¶

Where it fits in the stack¶

Typical use cases¶

Strengths¶

Limitations¶

When to use it¶

When not to use it¶

Getting started¶

Docker installation (Tika 3.0 Baseline)¶

Java Application (CLI)¶

Hello World¶

CLI examples¶

API examples¶

Python (using requests)¶

Recursive Metadata (curl)¶

n8n Integration: Automated PDF-to-Markdown¶

Workflow Pattern: Email Attachment to Markdown¶

Example Code Node (JS)¶

Sources / References¶

Backlog¶

Contribution Metadata¶

Apache Tika¶

What it is¶

What problem it solves¶

Where it fits in the stack¶

Typical use cases¶

Strengths¶

Limitations¶

When to use it¶

When not to use it¶

Getting started¶

Docker installation (Tika 3.0 Baseline)¶

Java Application (CLI)¶

Hello World¶

CLI examples¶

API examples¶

Python (using requests)¶

Recursive Metadata (curl)¶

n8n Integration: Automated PDF-to-Markdown¶

Workflow Pattern: Email Attachment to Markdown¶

Example Code Node (JS)¶

Related tools / concepts¶

Sources / References¶

Backlog¶

Contribution Metadata¶