GROBID Tools: Free PDF Metadata, DOI & Reference Extractor (In-Browser)

Looking for the GROBID library itself? This is an independent website and is not affiliated with the GROBID project. The open-source GROBID software lives at github.com/kermitt2/grobid with official documentation at grobid.readthedocs.io. Our guides explain what GROBID is and how to run it with Docker and its REST API.

📄

Drop PDF Here or Click to Choose

Processed locally in your browser. Maximum file size: 50MB

💡 Best results with text-based (born-digital) PDFs. Scanned or image-only PDFs need OCR first. This is a lightweight heuristic extractor; for large-scale, high-accuracy parsing see the open-source GROBID library.

What Gets Extracted Automatically

🔍

Bibliographic Metadata

Document title
Author names
Publication year
Abstract text
Keywords
Page count

🔗

Identifiers & Links

DOI (Digital Object Identifier)
URLs from references
ISBN numbers
arXiv IDs (old and new style)
PubMed IDs

📚

Citations & References

Full bibliography list
Citation parsing
Reference year detection
DOI extraction per citation
RIS, BibTeX and CSL-JSON export
Raw text preservation

Multiple Export Formats

Transform your PDF data into any format you need - all processed locally in your browser

📊

JSON

Structured data with all metadata & citations

📄

Plain Text

Clean UTF-8 text file (.txt)

📝

Word

Formatted DOCX document

📋

Markdown

GitHub-flavored Markdown (.md)

📚

BibTeX

Citation format for LaTeX (.bib)

🗂️

RIS

For EndNote, Zotero, Mendeley (.ris)

🧾

CSL-JSON

For Pandoc and CSL processors

Why Choose Browser-First PDF Tools?

🔒

100% Private

Your PDFs never leave your computer. All processing happens locally in your browser.

⚡

Lightning Fast

No upload and no server queue. PDF.js parses the file in a web worker on your own machine.

📡

Works Offline

After the first visit the app shell and libraries are cached, so the extractor keeps working without a connection.

🆓

Free to Use

No sign-up, no API limits, no upload quotas. Built on the open-source PDF.js library.

Learn more about PDF extraction

Practical guides on how PDF metadata is structured, how citation parsers actually work, and how to turn extracted records into a clean bibliography.

🧭

PDF metadata extraction

Where metadata lives inside a PDF, how heuristic extractors find titles and authors, and the failure modes worth knowing about.

📚

Citation parsing

How a parser splits a references section, labels each field, and keeps its hands off the ambiguous bits.

🔗

DOIs and identifiers

The persistent identifiers that make a citation resolvable — DOI, ArXiv, PubMed, ORCID, ISBN, ISSN.

🖨️

Scanned PDFs and OCR

What to do when a PDF has no text layer: tools, accuracy expectations, and privacy considerations.

🧩

Browser-side processing

The architectural choices behind a privacy-first PDF extractor, from PDF.js to service workers.

🧠

What is GROBID?

An independent overview of the open-source GROBID library: what it extracts, the TEI XML output, and where it is used.

🐳

Run GROBID with Docker

Start a local GROBID server, call the REST API with curl, and batch-process PDFs with the Python client.

⚖️

GROBID alternatives

CERMINE, AnyStyle, Nougat, Docling, Marker and browser tools compared for metadata and reference extraction.

🆔

arXiv identifiers

What an arXiv ID is, the old and new formats, versions, and how arXiv IDs relate to DOIs.

🔎

Find the DOI of a PDF

Where a DOI hides in a paper, and how to look one up from the title when the PDF does not print it.

🔁

BibTeX vs RIS

The practical differences between .bib, .ris, EndNote XML and CSL-JSON, and which to export.

🗂️

Export formats

Reference pages for BibTeX, RIS, CSL-JSON, Markdown, and DOCX, with notes on when to pick each.

Frequently asked questions

Is this the official GROBID website?

No. GROBID Tools is an independent site. The GROBID library is an open-source project developed on GitHub at github.com/kermitt2/grobid, with documentation at grobid.readthedocs.io. This site offers a separate, lightweight in-browser extractor and guides about GROBID and bibliographic formats.

Does the extractor upload my PDF?

No. The PDF is opened with PDF.js inside your browser and the extracted data is generated locally. The site uses Google Analytics for page-level visit statistics, but file names, file contents and extracted text are never sent anywhere.

Does this tool use the GROBID machine-learning models?

No. It uses lightweight heuristics (layout, font size and regular expressions) that run in the browser. GROBID itself is a Java service with trained models and gives more accurate, more structured results on large collections. See the guide on running GROBID with Docker.

Which export formats are available?

JSON, plain text, Microsoft Word (.docx), Markdown, BibTeX (.bib), RIS (.ris) and CSL-JSON. Reference strings are exported as found in the PDF, with any DOI, year and URL detected in them, so check them before citing.

Does it work with scanned PDFs?

Only if the PDF already has a text layer. Image-only scans need OCR first; the tool warns you when it finds almost no text.

GROBID Tools: PDF Metadata & Reference Extractor