CEUR-WS PDF Converter

Converts a Google Doc to a CEUR-WS–formatted PDF by downloading the document directly from its public URL, extracting front-matter content automatically, and compiling to PDF using either a full LaTeX pipeline (ceurart.cls) or LibreOffice as a fallback.

CEUR-WS Resources

• Submission instructions: https://ceur-ws.org/HOWTOSUBMIT.html
• CEURART styling package (LaTeX + LibreOffice templates): http://ceur-ws.org/Vol-XXX/CEURART.zip
  The ceurart.cls and elsarticle-num-names.bst files in this folder were extracted from that zip.
• ceur-precheck (validate your PDF before submission): https://github.com/johnbeve/ceur-precheck
• GenAI policy (required declaration for all submissions): https://ceur-ws.org/GenAI/Policy.html

Files

File	Purpose
ceur_pdf.py	Main conversion script
paper_config.json	Persisted paper metadata (authors, affiliations, conference, etc.)
ceurart.cls	CEUR-WS LaTeX document class (used by tectonic/pdflatex)
elsarticle-num-names.bst	BibTeX bibliography style (used when bib_file is set)

Quick Start

# Best output — install tectonic once (self-contained LaTeX, no MacTeX needed)
brew install tectonic

# Run from inside your paper folder — the script finds paper_config.json there
cd my_paper/
python3 /path/to/ceur_pdf.py -i "https://docs.google.com/document/d/<DOC_ID>"

# Subsequent runs: re-download and regenerate (still run from the paper folder)
python3 /path/to/ceur_pdf.py

# After editing authors, affiliations, ORCIDs, title or keywords in the doc:
python3 /path/to/ceur_pdf.py --extract   # or -e

# Or force the LibreOffice fallback (already installed on Mac)
python3 /path/to/ceur_pdf.py --engine lo

The script always operates on the current directory by default — run it from the folder that contains (or will contain) paper_config.json. Use -f <name> to target a subfolder of the current directory instead.

Note: a plain re-run only re-fetches the document body and abstract. Authors, affiliations, ORCIDs, title, and keywords are read from paper_config.json and will not update unless you pass --extract (-e).

Google Doc Layout

For the converter to auto-extract metadata, the document must open with a specific front-matter block before the first section heading. The simplest way to set this up once, then leave it — subsequent runs re-extract automatically.

Minimal front-matter template

How it looks in Google Docs (and how it renders as markdown):

The Full Paper Title Goes Here

Authors: Jane Smith^1,*, John Doe², Alice Brown^1,3

¹ First University, City, Country
² Second Institution, City, Country
³ Third Institution, City, Country

* Corresponding author jane.smith@example.com

ORCIDS:
JS: https://orcid.org/0000-0000-0000-0001
JD: https://orcid.org/0000-0000-0000-0002
AB: https://orcid.org/0000-0000-0000-0003

Keywords: food systems, ontology, processing

— # Abstract heading (Heading 1 style in Google Docs) —

Your abstract text goes here.

— # Introduction heading (Heading 1 style in Google Docs) —

Body text begins here …

The equivalent raw markdown syntax (what the converter sees after export):

The Full Paper Title Goes Here

Authors: Jane Smith^1,*^, John Doe^2^, Alice Brown^1,3^

^1^ First University, City, Country
^2^ Second Institution, City, Country
^3^ Third Institution, City, Country

* Corresponding author jane.smith@example.com

**ORCIDS:**
JS: https://orcid.org/0000-0000-0000-0001
JD: https://orcid.org/0000-0000-0000-0002
AB: https://orcid.org/0000-0000-0000-0003

**Keywords:** food systems, ontology, processing

# Abstract

Your abstract text goes here.

# Introduction

Body text begins here ...

Field-by-field rules

Title — the first non-blank paragraph that does not start with "Authors:". Plain text or Google Docs "Title" style both work.

Authors line — must start with Authors: (case-insensitive). List authors separated by commas. Each name is followed immediately by a superscript affiliation number (use Google Docs Insert → Special characters → Superscript, or type ^n^ notation). Mark the corresponding author with an additional * in the superscript, e.g. ^1,*^. Authors with multiple affiliations use comma-separated numbers: ^1,3^.

Affiliations — one per paragraph, in the form ^N^ Institution, City, Country where N matches the number used in the Authors line. Unicode superscript digits (¹ ² ³) are also recognised.

Corresponding author email — a paragraph containing the words "Corresponding author" followed by the email address (plain text or as a mailto: link).

ORCIDs — under a **ORCIDS:** bold heading (or any heading named "ORCIDs"), one line per author in the form XX: https://orcid.org/XXXX-... where XX is the author's initials. The converter matches initials to the author list; if two authors share initials, add an extra letter (e.g. JAS vs JAD).

Keywords — a bold inline label: **Keywords:** word, phrase, word (semicolons also work as separators). Place this anywhere in the front-matter block, typically after affiliations.

Abstract — a section headed # Abstract (any heading level, or a standalone **Abstract** bold paragraph). The converter strips this section from the body and places it in the formatted front matter automatically.

Section headings — use Google Docs "Heading 1" / "Heading 2" / "Heading 3" paragraph styles for \section / \subsection / \subsubsection in the output. Do not type heading numbers manually — ceurart.cls numbers sections automatically. Heading 4 is also supported (renders as \paragraph).

Tips

• Everything before the first # Heading is treated as front-matter; the converter ignores it in the body and extracts fields from it instead.
• Run python3 ceur_pdf.py --extract after restructuring the front-matter to force a re-parse and update paper_config.json.
• After the initial extraction you can hand-edit paper_config.json to correct anything (e.g. a misspelled affiliation) without touching the doc.
• The document sharing must be set to "Anyone with the link can view" in Google Docs — the converter accesses the export URL directly without signing in.
• Markdown as input: the converter currently only accepts Google Docs URLs. If your paper is already in a markdown file that follows the same front-matter conventions above, the conversion pipeline from the markdown step onward would work unchanged — local file input is a planned addition.

Re-running After Doc Changes

Just re-run the script — it always re-downloads the Google Doc from the live URL:

python3 ceur_pdf.py

With no -f flag the script looks for paper_config.json in the temp/ subfolder (relative to the script). Use -f <name> to point at a different folder, e.g. python3 ceur_pdf.py -f mypaper.

The abstract is re-extracted from the document on every run. Everything else (authors, affiliations, keywords, conference) comes from paper_config.json and stays stable between runs.

Note: Google's export endpoint can take up to ~5 minutes to reflect a recent edit. If your changes aren't showing up in the output, wait a few minutes and re-run.

To refresh all metadata from the current document (e.g., after adding authors or changing the title), use the --extract flag:

python3 ceur_pdf.py --extract

This re-parses the Google Doc front-matter and overwrites any auto-populated fields (title, authors, affiliations, ORCIDs, keywords) in paper_config.json. Fields you have hand-edited (conference, copyright, GenAI declaration, etc.) are not touched unless they were originally extracted from the doc.

Output Quality

Engine	How to get it	Output
tectonic	brew install tectonic	Full CEURART layout: author blocks with ORCIDs, conference footer, copyright line, CC-BY icon
LibreOffice	Pre-installed on Mac	Converts the Google Doc's own formatting to PDF — no CEURART layout applied

Install tectonic for submission-ready output.

What Is Auto-Populated

When the script runs it prints a full summary of the metadata being used. Things extracted automatically from the Google Doc on each run:

• Abstract — pulled from the # Abstract section heading in the document body

Things stored in paper_config.json (set once, reused on every run):

• Title — set manually; update if the doc title changes
• Authors — names, affiliation numbers, ORCIDs, email, corresponding-author flag
• Affiliations — numbered institution names matching the superscripts in the doc
• Keywords — keyword list
• Conference — workshop name, date, location
• Copyright year / clause — defaults to CC-BY 4.0
• Acknowledgments, GenAI declaration, bibliography file

paper_config.json Reference

{
  "doc_url": "",                     // set via -i <url> or edit directly
  "export_format": "docx",          // docx recommended; odt also works
                                     // (other document types planned for future support)
  "output_pdf": "my_paper.pdf",

  "pdf_engine": "auto",             // auto | tex | lo

  "title": "Paper Title Here",
  "title_note": "",                 // optional footnote on the title (★ mark)
  "conference": "Workshop Name, Month DD-DD, YYYY, City, Country",
  "copyrightyear": "2026",
  "copyrightclause": "Copyright for this paper by its authors. ...",

  "abstract": "",                   // leave empty to extract from doc automatically
  "abstract_from_doc": true,        // set false to use the abstract field above instead
  "abstract_heading": "Abstract",   // heading text that marks the abstract section

  "keywords": ["keyword1", "keyword2"],
  "keywords_heading": "Keywords",   // heading text to strip from body

  "authors": [
    {
      "name": "Full Name",
      "affils": "1",                // comma-separated affiliation numbers, e.g. "1,2"
      "orcid": "0000-0000-0000-0000",
      "email": "name@example.com",
      "url": "",
      "corresponding": true,
      "equal_contribution": false
    }
  ],

  "affiliations": [
    {"num": "1", "text": "Institution, Address, City, Country"}
  ],

  "cortext": "Corresponding author.",
  "acknowledgments": "",

  // GenAI declaration — see https://ceur-ws.org/GenAI/Policy.html
  // If all three fields are empty the script prompts interactively on first run.
  "genai_uses": [],                 // subset of: text_creation, translation, grammar,
                                    //   rephrasing, image_creation
  "genai_tools": "",                // free text, e.g. "ChatGPT, Grammarly"
  "genai_declaration": "",          // auto-composed from uses+tools; override here if needed

  "bib_file": "",                   // basename of a .bib file (no extension) if using BibTeX

  "pandoc_extra_args": []           // extra flags passed to pandoc, e.g. ["--csl=apa.csl"]
}

Supported Input Formats

Currently the script accepts Google Docs URLs (exported as DOCX via the Google Docs API). Support for additional document types (e.g. local DOCX/ODT files, Overleaf exports, Markdown) is planned for a future version.

GenAI Declaration

CEUR-WS requires a Generative AI use statement in every paper:

• Policy: https://ceur-ws.org/GenAI/Policy.html
• Full use-case taxonomy: https://ceur-ws.org/GenAI/Taxonomy.html

The script handles this automatically:

• First run (when genai_declaration is empty): the script prompts you interactively to select any applicable use cases and name the tool(s) used, then saves the composed text back to paper_config.json.
• Subsequent runs: the saved declaration is used as-is.
• No AI used: just press Enter at the prompt — the standard "no tools used" statement is written automatically.
• Manual override: set genai_declaration to any text you like and the prompt is skipped entirely.

Valid use-case keys for genai_uses (drawn from the CEUR-WS taxonomy):

Key	Label
text_creation	Drafting Content
image_creation	Generate Images
translation	Text Translation
literature_review	Generate Literature Review
rephrasing	Paraphrase and Reword
improve_style	Improve Writing Style
abstract_drafting	Abstract Drafting
grammar	Grammar and Spelling Check
plagiarism_detection	Plagiarism Detection
citation_management	Citation Management
formatting	Formatting Assistance
peer_review	Peer Review Simulation
content_enhancement	Content Enhancement

Command-Line Options

python3 ceur_pdf.py [options]

  -i, --input URL    Google Doc URL — required when paper_config.json does not exist
                     or doc_url is empty; updates doc_url if a different URL is given
  -f, --folder DIR   Working folder for paper_config.json, output PDF, and paper.tex
                     (relative to current directory; default: current directory)
  --engine auto      Auto-detect: prefer LaTeX (tectonic/pdflatex), fall back to LibreOffice
  --engine tex       Force LaTeX (exits with error if no engine found)
  --engine lo        Force LibreOffice
  -e, --extract      Re-extract metadata (title, authors, affiliations, keywords) from
                     the current Google Doc, overwriting any previously auto-populated
                     values in paper_config.json
  --keep-tex         Save the intermediate paper.tex in the working folder for inspection
  --settings FILE    Override the settings JSON path (default: <folder>/paper_config.json)

By default the script operates on the current directory:

my_paper/
  paper_config.json   ← paper metadata (created on first -i run)
  paper.pdf           ← final output PDF
  paper.tex           ← intermediate LaTeX (only with --keep-tex)

To manage multiple papers, keep each in its own folder and cd into it:

# Set up two separate papers
cd paper_a/  && python3 /path/to/ceur_pdf.py -i "https://docs.google.com/document/d/<ID_A>"
cd ../paper_b/ && python3 /path/to/ceur_pdf.py -i "https://docs.google.com/document/d/<ID_B>"

# Regenerate each independently
cd paper_a/ && python3 /path/to/ceur_pdf.py
cd paper_b/ && python3 /path/to/ceur_pdf.py

Or use -f from a common parent directory:

python3 /path/to/ceur_pdf.py -f paper_a
python3 /path/to/ceur_pdf.py -f paper_b