Skill: Book Creation in PaperLab

When to Use

USE WHEN: creating a new scientific book project, writing chapter content, running book notebooks, building/rendering a book, or troubleshooting the book pipeline. This skill covers the complete lifecycle from book-new to final rendered outputs in HTML/Word/PDF/ODF.

DO NOT USE for: paper writing workflows (use write-methods-section, journal-formatting, etc. instead).

1. Book Project Structure

Every book lives under neqsim-paperlab/books/<slug>_<year>/. The critical file is book.yaml — it defines metadata, chapter ordering, parts, and rendering settings.

books/<book_slug>/
├── book.yaml                  # Master manifest — defines EVERYTHING
├── refs.bib                   # Master bibliography
├── nomenclature.yaml          # Symbol definitions
├── frontmatter/
│   ├── title_page.md
│   ├── copyright.md
│   ├── dedication.md
│   └── preface.md
├── chapters/
│   ├── ch01_introduction/
│   │   ├── chapter.md         # Chapter text (markdown with LaTeX math)
│   │   ├── figures/           # Generated figures (PNG)
│   │   └── notebooks/         # Jupyter notebooks that generate figures
│   ├── ch02_topic_name/
│   │   ├── chapter.md
│   │   ├── figures/
│   │   └── notebooks/
│   └── ...
├── backmatter/
│   ├── glossary.md
│   └── author_bio.md
└── submission/                # Rendered outputs
    ├── book.html
    ├── book.docx
    ├── book.pdf
   ├── book.odt
    └── figures_chNN/          # PDF renderer copies figures here for Typst

book.yaml Structure

title: "Book Title"
subtitle: "Subtitle"
authors:
  - name: "Author Name"
    affiliation: "Institution"
    email: "email@example.com"
edition: "1st"
year: 2026
publisher: "springer"        # springer | wiley | crc | self
language: "en"

settings:
  book_type: "textbook"      # simple | technical | textbook (see §1.1)
  page_size: "b5"            # b5 | a4
  font_size: 10
  line_spacing: 1.2
  two_sided: true
  chapter_numbering: true
  equation_numbering: "chapter"  # chapter-scoped: Eq. 3.1, 3.2, ...
  # Optional content toggles (override the book_type profile, see §1.1):
  # include_copyright: true
  # include_learning_objectives: true
  # include_exercises: true
  # exclude_sections: ["Review Questions", "Problems"]

frontmatter:
  - title_page
  - copyright
  - dedication
  - preface

parts:
  - title: "Part I: Foundations"
    chapters:
      - dir: "ch01_introduction"           # MUST match exact dir name
        title: "Introduction and Context"
      - dir: "ch02_topic_name"
        title: "Topic Name"
  - title: "Part II: Advanced"
    chapters:
      - dir: "ch03_another_topic"
        title: "Another Topic"

backmatter:
  - glossary
  - author_bio

nomenclature:
  file: "nomenclature.yaml"
  position: "after_toc"

bibliography:
  style: "numeric"
  file: "refs.bib"

Critical Rules for book.yaml

1. dir values MUST match actual directory names under chapters/. Mismatches cause the renderer to silently skip chapters.
2. Chapter numbering is sequential across all parts — the renderers count 1, 2, 3, ... regardless of part boundaries.
3. Do not create alternate directory names (e.g., ch03_cubic_eos vs ch03_cubic_equations_of_state). Only the dir listed in book.yaml is used. Remove unused dirs to avoid confusion.
4. Publisher profiles in books/_publisher_profiles/ define page size, margins, and fonts. The publisher key selects which profile to load.

Optional metadata

The following top-level keys are optional. Omit or leave them blank and the renderers degrade gracefully (no author line, no publisher imprint, etc.): authors, publisher, year, language, isbn.

1.1 Book Type Profiles (content level)

settings.book_type selects how much pedagogical scaffolding is included so the same manuscript can be published at three different levels. It controls whether the copyright page, learning objectives, exercises, and review sections are rendered.

• Default is textbook (everything included). Omitting book_type reproduces the historic full-content behaviour, so existing books are unaffected.
• The profile decides which ## <Heading> sections are stripped from each chapter.md at render time. The chapter source files are never modified — authors can keep all sections in the source and switch the output level from book.yaml alone.

Overriding a profile

Explicit settings.include_* and settings.exclude_sections keys always win over the book_type profile. This lets you start from a profile and tweak a single aspect:

settings:
  book_type: technical        # baseline: no pedagogy, keeps copyright
  include_copyright: false    # …but drop the copyright page anyway

settings:
  book_type: simple                 # baseline: strip everything
  include_learning_objectives: true # …but keep Learning Objectives

Resolution order for each toggle: explicit settings key → book_type profile default → built-in default (include). Accepted boolean strings are case-insensitive: true/yes/1/on and false/no/0/off.

exclude_sections is additive — the titles you list are removed in addition to whatever the profile already strips. Matching is case-insensitive and ignores leading numbers (## 5.3 Problems matches Problems).

2. Creating a New Book

Command

cd neqsim-paperlab
python paperflow.py book-new "Book Title" --publisher springer --chapters 12

This creates the scaffold with empty chapters (ch01 through ch12).

Use --type to pick the content level (writes settings.book_type into the generated book.yaml, see §1.1). Defaults to textbook:

python paperflow.py book-new "Field Handbook" --type technical --chapters 8
python paperflow.py book-new "Quick Reference" --type simple --chapters 4

Post-Scaffold Steps (MANDATORY)

1. Rename chapter directories to descriptive names:

   ch01 → ch01_introduction
   ch02 → ch02_thermodynamic_foundations

2. Update book.yaml to match the new directory names and set real titles.
3. Organize into parts by editing the parts: section in book.yaml.
4. Fill frontmatter — at minimum: title_page.md (title, authors), preface.md (motivation, audience, acknowledgements).

5. Deep Literature Review and refs.bib (DO THIS BEFORE WRITING CHAPTERS):

   This step is non-negotiable — a book with sparse references signals shallow scholarship. Complete the bibliography BEFORE writing chapter content.

   a. Build master refs.bib at the book root with 100+ entries minimum. Organize into clearly commented sections by topic:

   % ─── Foundational thermodynamics ───
   @book{Prausnitz1999, ... }
   
   % ─── Equations of state ───
   @article{Soave1972, ... }

   b. Mine existing PaperLab papers — Search papers/*/refs.bib for related citations. Reuse BibTeX entries for consistency:

   grep -rl "keyword" papers/*/refs.bib

   c. Ensure coverage per chapter — Each chapter should have 10–20+ unique references identified before writing begins. d. Include all categories: seminal/foundational works, recent advances (last 5 years), textbooks, experimental data sources, competing methods, review articles.

   See PAPER_WRITING_GUIDELINES.md "MANDATORY: Literature Review and Citation Collection First" for the complete workflow and quality rules.

6. Import content from PaperLab papers (MANDATORY) — Before writing each chapter, check papers/ for papers that cover the same topic:

   grep -rl "chapter_keyword" papers/*/paper.md papers/*/plan.json

   For each matching paper:

   • Copy figures from papers/<paper>/figures/*.png to chapters/chNN/figures/
   • Copy tables (markdown) from paper.md into chapter.md
   • Incorporate results — numerical benchmarks, validation data, performance comparisons
   • Adapt text — expand paper explanations for the book's broader audience
   • Reuse equations — copy LaTeX from paper.md, ensure notation matches book nomenclature

   This step ensures the book benefits from the detailed research already done in PaperLab papers. See PAPER_WRITING_GUIDELINES.md for the full workflow.

7. Create chapter content — each chapter.md uses markdown with:
   • $$...$$ for display equations (rendered as OMML in Word, KaTeX in HTML)
   • $...$ for inline math
   • \cite{key} tags for every claim, equation origin, and data source
   • ![Caption](figures/filename.png) for figures
   • Standard markdown tables (rendered as booktabs-style)

3. Writing Chapter Content

Equations

Display equations use $$...$$ delimiters. They are automatically numbered per-chapter (Eq. 1.1, 1.2, ..., Eq. 2.1, 2.2, ...).

The cubic equation of state is:

$$
P = \frac{RT}{V_m - b} - \frac{a(T)}{(V_m + \epsilon b)(V_m + \sigma b)}
$$

For the SRK model, the alpha function is:

$$
\alpha(T) = \left[1 + m\left(1 - \sqrt{T_r}\right)\right]^2
$$

Inline math uses $...$:

The acentric factor $\omega$ affects the alpha function $\alpha(T_r, \omega)$.

Equation Pipeline (per renderer)

Word renderer dependency: Requires MML2OMML.XSL from Microsoft Office. Location: C:\Program Files\Microsoft Office\root\Office16\MML2OMML.XSL. The math_utils.py module handles the conversion pipeline.

Figures

Figures are generated by Jupyter notebooks in chapters/chNN/notebooks/ and saved to chapters/chNN/figures/. Reference them in chapter.md as:

![Phase diagram showing vapor-liquid equilibrium](figures/vle_phase_diagram.png)

The renderers handle figure numbering automatically (Figure 1.1, 1.2, ...).

Tables

Use standard markdown tables. The Word renderer applies booktabs-style formatting (horizontal rules only, no vertical lines):

| Component | Tc (K) | Pc (kPa) | ω |
|-----------|--------|----------|------|
| Methane   | 190.56 | 4599     | 0.0115 |
| Ethane    | 305.32 | 4872     | 0.0995 |

Code Blocks

Use fenced code blocks with language specifiers:

```python
from neqsim import jneqsim
fluid = jneqsim.thermo.system.SystemSrkEos(298.15, 50.0)
```

4. Jupyter Notebooks

Each chapter has a notebooks/ directory containing .ipynb files that generate the chapter's figures.

Notebook Setup Cell (MANDATORY first cell)

import importlib, subprocess, sys

try:
    from neqsim_dev_setup import neqsim_init, neqsim_classes
    ns = neqsim_init(recompile=False)
    ns = neqsim_classes(ns)
    NEQSIM_MODE = "devtools"
    print("NeqSim loaded via devtools (local dev mode)")
except Exception:
    try:
        import neqsim
    except ImportError:
        subprocess.check_call([sys.executable, "-m", "pip", "install", "-q", "neqsim"])
    from neqsim import jneqsim
    NEQSIM_MODE = "pip"
    print("NeqSim loaded via pip package")

Saving Figures (MANDATORY pattern)

import matplotlib.pyplot as plt
from pathlib import Path

# Resolve to chapter figures/ directory
NOTEBOOK_DIR = Path(globals().get(
    "__vsc_ipynb_file__", __file__
)).resolve().parent
FIGURES_DIR = NOTEBOOK_DIR.parent / "figures"
FIGURES_DIR.mkdir(exist_ok=True)

fig, ax = plt.subplots(figsize=(6, 4))
# ... plotting code ...
fig.savefig(FIGURES_DIR / "figure_name.png", dpi=150, bbox_inches="tight")
plt.show()

Running Notebooks

# Run all book notebooks (compiles NeqSim first)
python paperflow.py book-run-notebooks books/<book_dir>

# Run specific chapter's notebooks
python paperflow.py book-run-notebooks books/<book_dir> --chapter ch04_association_theory

# Skip compilation if already up to date
python paperflow.py book-run-notebooks books/<book_dir> --no-compile

# Full build (compile + notebooks + check + render)
python paperflow.py book-build books/<book_dir> --format all

5. Building and Rendering

Build Commands

cd neqsim-paperlab

# Full build: compile → notebooks → quality checks → render all formats
python paperflow.py book-build books/<book_dir> --format all

# Skip notebooks (use existing figure outputs)
python paperflow.py book-build books/<book_dir> --format all --skip-notebooks

# Skip compilation (notebooks still run using existing classes)
python paperflow.py book-build books/<book_dir> --format all --no-compile

# Skip both (render only — fastest for formatting iterations)
python paperflow.py book-build books/<book_dir> --format all --skip-notebooks --no-compile

# Single format
python paperflow.py book-render books/<book_dir> --format docx
python paperflow.py book-render books/<book_dir> --format html
python paperflow.py book-render books/<book_dir> --format pdf
python paperflow.py book-render books/<book_dir> --format odf

Build Pipeline Steps

The book-build command runs these steps in order:

1. Compile — mvnw compile (ensures Java classes are current)
2. Notebooks — execute all .ipynb files, capturing errors
3. Figure injection — auto-insert new figures into chapter.md
4. Quality checks — structure, completeness, consistency
5. Render — produce output in requested format(s)

Output Locations

PDF Figure Handling

The PDF renderer copies chapter figures to submission/figures_chNN/ because Typst needs them alongside the master .typ file. This is by design — the copies are build artifacts in submission/ and should not be committed.

6. Quality Checks

python paperflow.py book-check books/<book_dir>

Checks: structure (dirs exist), completeness (word counts), consistency (cross-references), bibliography, figure references, nomenclature.

7. Common Issues and Solutions

Issue: Chapter silently skipped in render

Cause: The dir value in book.yaml doesn't match the actual directory name. Fix: Verify ls chapters/ matches every dir: entry in book.yaml.

Issue: Equations render as plain text in Word

Cause: MML2OMML.XSL not found (no Office installed) or latex2mathml not installed. Fix: Install Office or ensure MML2OMML.XSL is on the system. Install dependencies: pip install latex2mathml lxml.

Issue: Equations render as Unicode instead of proper math in ODF

Expected: ODF renderer uses Unicode Greek/subscript/superscript as fallback. Full equation rendering in ODF requires a dedicated math library.

Issue: Duplicate chapter directories

Cause: Chapters renamed but old directories not deleted. Fix: Only the dir listed in book.yaml is used. Delete unused directories.

Issue: Stale Python module imports

Cause: __pycache__ holds old versions of modified tools. Fix: Delete tools/__pycache__/ and re-run.

Issue: Figures not appearing in rendered output

Cause: Figure path in chapter.md doesn't match actual filename. Fix: Use ![Caption](figures/exact_filename.png) — paths are relative to the chapter directory.

Issue: PDF build fails

Cause: Missing pandoc or typst package. Fix: Install pandoc (system package) and pip install typst.

Issue: Typst PDF compile fails with unexpected underscore

Cause: Bibliography entries from refs.bib contain LaTeX-style fragments such as $_2$, {CO, or \textit{...}. When inserted into a Typst content block these $...$ segments are read as math mode, where a leading _ has no base. Fix: The bibliography renderer in tools/book_render_pdf.py escapes \ # $ { } _ * @ < > [ ] \``. Do not remove that escape table. If you add a new bibliography style, make sure the entry text is escaped before being embedded into a #par[...]` block.

Issue: Typst PDF compile fails with expected alignment or auto, found string

Cause: #set page(binding: "left") no longer accepts a string in Typst 0.14. The keyword expects an alignment value (left / right). Fix: Use binding: left (no quotes) in the generated preamble.

7a. Professional Typesetting

The PDF renderer (tools/book_render_pdf.py, function build_book_typst_preamble) emits a Typst preamble that produces a publisher-quality book layout. When you modify the preamble, preserve the following features and verify against the checklist below.

Layout Features Built In

Typesetting Quality Checklist (run before final PDF render)

1. Publisher profile (books/_publisher_profiles/<name>.yaml) exists for the value of publisher: in book.yaml. If missing, the renderer falls back to A4 / generic margins — set the profile explicitly.
2. trim_size follows the format "WxHmm" (e.g. "155x235mm").
3. font is installed on the host (default "New Computer Modern"). On Windows, install via the New Computer Modern release zip.
4. After running book-render --format pdf, open the PDF and verify:
   • Chapter 1 starts on a recto (right-hand) page.
   • Even-page header shows the chapter title left-aligned; odd-page header shows it right-aligned; both in smallcaps.
   • Page 1 (title page) has no running header.
   • Each chapter opener shows the smallcaps "Chapter N" line, a thin coloured rule, and the chapter title set in the heading colour.
   • Equation labels read (1.1), (2.1), ... and reset at every chapter.
   • Tables show only thin top + bottom rules (booktabs style); no vertical rules.
   • Reference list is indented hanging-style, with bold reference numbers.
   • Body text is justified with hyphenation; no rivers or large gaps.

5. Run the consistency checker:

   python paperflow.py book-check books/<book_dir>

6. The generated submission/book.typ is a build artifact. Do not edit it by hand — change tools/book_render_pdf.py and re-render.

Modifying the Preamble Safely

If you add new Typst rules to build_book_typst_preamble:

1. The function returns an f-string; literal { and } must be doubled ({{ / }}).
2. Test with the smallest book first: python paperflow.py book-render books/<small_book> --format pdf.
3. If compilation fails with a misleading message (e.g. unexpected underscore from a bib entry), bisect by chapter — write _test.typ inside submission/ so figure paths still resolve.
4. Typst 0.14 syntax notes:
   • binding: takes an alignment, not a string.
   • #set page(width: ..., height: ...) uses keyword args, not a tuple.
   • pagebreak(to: ...) accepts "odd" / "even" strings or omit the arg.
   • Use query(heading.where(level: 1).before(here())) for running headers (avoid shadowing the built-in here function with a local variable).

8. Workflow for Iterative Chapter Writing

The fastest iteration loop for writing/editing chapters:

1. Edit chapter.md (add equations, text, figure references)
2. Render single format for preview:

   python paperflow.py book-build books/<book_dir> --format html --skip-notebooks --no-compile

3. Open submission/book.html in browser — instant feedback
4. When satisfied, do a full build:

   python paperflow.py book-build books/<book_dir> --format all --skip-notebooks

For notebook development:

1. Edit notebooks in VS Code / JupyterLab
2. Run individual notebooks manually to iterate on figures
3. When figures are ready, rebuild:

   python paperflow.py book-build books/<book_dir> --format all --skip-notebooks --no-compile

9. Tools Reference

9a. Live preview while writing

python paperflow.py book-preview books/<book_dir>
# opens http://127.0.0.1:8765 ; auto-reloads on every chapter.md or book.yaml save

9b. CSL citations

Add to book.yaml:

bibliography:
  file: refs.bib
  csl: styles/elsevier-harvard.csl   # any CSL 1.0.2 file

The PDF, HTML and EPUB renderers will route citations through Pandoc's --citeproc engine when both files exist.

9c. EPUB output

python paperflow.py book-render books/<book_dir> --format epub
# requires pandoc on PATH; reads book.yaml metadata + cover.png if present

9d. Crossref bibliography enrichment

python paperflow.py book-enrich-bib books/<book_dir>            # writes refs.enriched.bib
python paperflow.py book-enrich-bib books/<book_dir> --in-place # overwrites refs.bib

Validates every DOI, warns on title/year mismatches, and adds DOIs to entries that don't have one when Crossref's top match exceeds the threshold.

9e. PDF/X-1a or PDF/A conversion

If a publisher profile sets:

print_specs:
  pdf_standard: PDF/X-1a:2003
  color_profile: CMYK

then the PDF renderer automatically post-processes book.pdf with Ghostscript. The original is kept as book.pre-pdfx.pdf.

9f. NeqSim integration for scientific writing

For chapters that defend numbers via NeqSim simulation, load the companion skill skills/neqsim-in-writing/SKILL.md. It covers the dual-boot setup cell, claim-to-test linkage (@neqsim:claim), equation-to-Java method cross-references (@neqsim:eq), units enforcement against nomenclature.yaml, and notebook-driven figure / results-table injection (@neqsim:figure, @neqsim:table).

10. Full End-to-End Example

# 1. Create project
cd neqsim-paperlab
python paperflow.py book-new "Thermodynamic Modeling" --publisher springer --chapters 8

# 2. Edit book.yaml — rename chapters, set titles, organize parts

# 3. Write chapter content in chapters/ch01_xxx/chapter.md

# 4. Create notebooks in chapters/ch01_xxx/notebooks/

# 5. Run notebooks to generate figures
python paperflow.py book-run-notebooks books/thermodynamic_modeling_2026/

# 6. Build all formats
python paperflow.py book-build books/thermodynamic_modeling_2026/ --format all

# 7. Check quality
python paperflow.py book-check books/thermodynamic_modeling_2026/

# 8. Iterate — edit, rebuild, review
python paperflow.py book-build books/thermodynamic_modeling_2026/ --format html --skip-notebooks --no-compile