docu-craft 0.4.0

Convert Markdown to polished documents with reusable themes

docu-craft

Stop writing PDFs by hand. Write Markdown, get a polished document.

import docu_craft

docu_craft.render("report.md", theme="scholar")

---

What it does

You write content in Markdown. docu-craft handles everything else:

• Picks a theme — fonts, colors, margins, tables, code blocks, all styled consistently across every output format
• Named document styles — styles are registered as first-class named styles in DOCX and ODT, visible in Word's Styles pane and LibreOffice's Styles panel, not just inline formatting
• Multiple output formats — PDF, HTML, DOCX, ODT, LaTeX from the same source file
• Handles emoji — system font fallback or image replacement via downloadable emoji sets
• Validates structure — checks that your document has the sections it needs before rendering
• Finds your assets — themes, skeletons, and emoji sets from the package, your home folder, or any mounted path

No more pasting CSS into scripts. No more fighting with PDF libraries.

---

Install

pip install docu-craft

# with DOCX support
pip install "docu-craft[docx]"

# with HTML → Markdown conversion
pip install "docu-craft[html]"

# with PDF → Markdown extraction
pip install "docu-craft[pymupdf]"

# everything
pip install "docu-craft[all]"

---

Conversion paths

docu-craft uses a weighted DAG to resolve conversions. Ask for any path — it finds the route automatically.

From	To	Engine	Notes
md	pdf	WeasyPrint (default)	via HTML — full CSS, emoji font support
md	pdf	LaTeX	via pdflatex/xelatex — requires LaTeX install
md	html	—	embedded CSS from theme
md	docx	—	named styles visible in Word's Styles pane
md	odt	—	named styles visible in LibreOffice's Styles panel
md	latex	—	direct Markdown → LaTeX source
html	md	—	extracts article body, strips chrome, preserves images
pdf	md	—	extracts structured text, infers headings from font size

doc = docu_craft.Document("report.md")
doc.apply_theme("scholar")

doc.render(format="pdf",  output="report.pdf")
doc.render(format="docx", output="report.docx")
doc.render(format="html", output="report.html")

# Convert a web paper to Markdown
doc = docu_craft.Document("paper.html")
doc.render(format="md", output="paper.md", img_dir="figures/", base_url="https://example.com/paper/")

# Extract text from a PDF
doc = docu_craft.Document("paper.pdf")
doc.render(format="md", output="paper.md")

---

Themes

Themes define the full visual identity of a document — fonts, colors, spacing, and every named style — applied consistently across all output formats.

Theme	Best for
scholar	Academic articles, PhD documents
handout	Course materials, workshops
tech-doc	API docs, technical references
official	Institutional letters, formal reports

docu_craft.render("thesis.md", theme="scholar")
docu_craft.render("class_notes.md", theme="handout")

Drop your own theme in ~/docu_craft/themes/mytheme/ and use it the same way.

Theme schema

Every theme defines a styles block with named semantic styles — body, heading1–heading6, code_block, code_inline, table_header, table_cell, list_item, quote. Each style references a font stack by name (body, header, or mono) and specifies size, color, weight, spacing, and background.

# theme.yaml
style:
  fonts:
    body:   ["Georgia", "Times New Roman", "serif"]
    header: ["Arial", "Helvetica", "sans-serif"]
    mono:   ["Courier New", "Courier", "monospace"]
    emoji:  ["Noto Color Emoji", "Apple Color Emoji", "Segoe UI Emoji"]

  styles:
    heading1:
      font: header
      size: 16
      color: "#1a1a2e"
      bold: true
    code_block:
      font: mono
      size: 9
      background: "#f4f4f4"

The same style definitions drive all renderers — Word's "DC Heading 1", LibreOffice's "DC Body Text", and the CSS h1 rule all come from the same source.

---

Emoji

System emoji (font fallback)

By default docu-craft passes emoji through and lets the output application render them using its own font stack. For PDF via WeasyPrint, emoji fonts are resolved automatically:

# theme.yaml
style:
  fonts:
    emoji: ["Noto Color Emoji", "Apple Color Emoji", "Segoe UI Emoji", "Twemoji Mozilla"]

docu-craft probes the filesystem for each font (in order: Linux → macOS → Windows) and emits @font-face rules pointing to the actual font files. On WSL2, it finds Windows fonts via /mnt/c/Windows/Fonts/ automatically.

Custom emoji sets (image replacement)

For consistent, platform-independent emoji across all readers, use an image-based emoji set. Each emoji character is replaced with a PNG from the set.

docu_craft.render("notes.md", theme="scholar", emoji_set="twemoji")

Download sets with the built-in downloader:

python -m docu_craft.emoji.downloader twemoji
python -m docu_craft.emoji.downloader noto

Twemoji

The recommended custom set. Twemoji (CC-BY 4.0, by Twitter/jdecked) draws every glyph on the same grid at the same size with the same line weights. When you place several emoji side by side they read as a unified visual system — consistent weight, consistent optical size, no surprises. This is the same discipline a monospaced font applies to letterforms. If visual consistency across emoji matters in your documents, Twemoji is the right choice.

Other available sets:

Set	License	Coverage	Character
twemoji	CC-BY 4.0	3800+	Grid-locked, uniform, consistent
noto	Apache 2.0	3000+	Disciplined, clean, Google design system

---

Skeletons

Skeletons define what sections a document should have. docu-craft tells you if something is missing before you render.

doc = docu_craft.Document("thesis.md")
doc.apply_skeleton("academic_article").validate()
doc.render()

Built-in skeletons: academic_article, plan_trabajo, tech_doc, official_letter, course_handout.

---

Config layers

Set defaults once, override whenever you need:

# ~/docu_craft/config.yaml — your personal defaults
defaults:
  theme: scholar
  emoji_set: twemoji

# .docu_craft.yaml — per-project overrides
defaults:
  theme: handout

---
theme: official
---
# Per-document frontmatter overrides everything above

# Explicit argument wins over all of the above
docu_craft.render("file.md", theme="tech-doc")

---

Extended storage

Point docu-craft at any folder — a team share, a mounted drive, a network path. It searches all of them for themes, skeletons, and emoji sets.

docu_craft.add_extended_store("/mnt/team/docu_craft-assets", name="team")

# .docu_craft.yaml
extended_stores:
  - /mnt/team/docu_craft-assets
  - path: G:/Shared/styles
    name: gdrive

---

Pluggable renderers

docu_craft.register_renderer(
    format="pdf",
    module_path="mypackage.renderer:MyRenderer",
    engine="myengine",
    package="mypackage",
    install="pip install mypackage",
)

doc.render(format="pdf", engine="myengine")

The renderer graph is a weighted DAG — preference between equivalent paths (e.g. md→html→pdf vs md→latex→pdf) is configured per project via the engine setting.

---

Demo

The demo/ folder contains a ready-to-run example:

cd demo
python render.py

render.py exercises every major conversion path — PDF, HTML, DOCX, ODT, LaTeX, and the html→md and pdf→md extractors. Output files land in demo/ alongside the script.

---

License

Apache 2.0 © Christian A. Servin Lozano