mark-everything 0.1.0

Convert any document (.docx, .pptx, .xlsx, .eml, .msg, .pdf) to clean Markdown.

MarkEverything

Convert any Office document or email into clean, structured Markdown — preserving headings, tables, images, diagrams, and meaning.

Format	What is extracted
.docx	Headings, paragraphs, bold/italic, tables, hyperlinks, embedded images, SmartArt → Mermaid, OML equations → LaTeX
.pptx	Per-slide sections, spatial reading order, connected shapes → Mermaid, embedded images, speaker notes
.xlsx	Sheet tables (used-range only), bold cells, Bar/Pie charts → Mermaid
.eml	YAML frontmatter headers, HTML/plain body, inline images, attachment list
.msg	Same as .eml via extract-msg

---

Requirements

• Python 3.10+
• See requirements.txt for library dependencies

---

Installation

From source (development)

git clone https://github.com/your-org/mark-everything.git
cd mark-everything
pip install -e ".[dev]"

Runtime only

pip install -e .

With .msg support

pip install -e ".[msg]"

---

Quick Start

Python API

from mark_everything import convert
from mark_everything.models import ConversionConfig, ImageMode, MermaidDepth

# Minimal — all defaults (images embedded as base64)
result = convert("report.docx")
print(result.markdown)

# Full config
config = ConversionConfig(
    image_mode=ImageMode.PATH,       # save images to disk
    output_dir="output/assets",      # where to save them
    mermaid_depth=MermaidDepth.FULL, # attempt complex SmartArt
    exclude_sheets=["Internal"],     # skip this Excel sheet
    include_speaker_notes=True,      # include PPT speaker notes
)
result = convert("presentation.pptx", config)

# Inspect result
print(result.markdown)           # the Markdown string
print(result.asset_paths)        # list[Path] of saved images
print(result.warnings)           # list[str] of non-fatal issues

Command-Line Interface

# Basic — print Markdown to stdout
mark-everything report.docx

# Write to file
mark-everything report.docx -o output/report.md

# Save images as files instead of base64
mark-everything report.docx --image-mode path --output-dir output/assets -o output/report.md

# PowerPoint — omit speaker notes
mark-everything deck.pptx --no-speaker-notes -o output/deck.md

# Excel — skip certain sheets, full SmartArt depth
mark-everything data.xlsx --exclude-sheets Internal Scratch --mermaid-depth full -o output/data.md

# Verbose logging
mark-everything email.eml -v -o output/email.md

---

Configuration Reference

Option	Type	Default	Description
image_mode	"base64" | "path"	"base64"	Embed images as data-URIs or save to disk
output_dir	Path | None	None	Asset directory when image_mode="path"
mermaid_depth	"simple" | "full"	"simple"	SmartArt / connected-shape conversion depth
exclude_sheets	list[str]	[]	Excel sheet names to skip
include_speaker_notes	bool	True	Include PPT speaker notes as > blockquotes

---

Project Structure

mark_everything/
├── __init__.py          # Public API: convert()
├── engine.py            # Central router
├── models.py            # Pydantic config + result models
├── media.py             # Image encode / save helpers
├── mermaid.py           # Mermaid diagram builder
├── cli.py               # CLI entry-point
└── converters/
    ├── docx.py          # Word converter
    ├── pptx.py          # PowerPoint converter
    ├── xlsx.py          # Excel converter
    └── email.py         # Email (.eml / .msg) converter

---

Design Principles

• Pure functions — all helpers are stateless; same input always yields same output.
• Pydantic models — ConversionConfig is validated and frozen; ConversionResult is immutable.
• Structured logging — logging.getLogger(__name__) throughout; no print() in library code.
• No memory leaks — all file handles and workbooks are opened inside with / try-finally blocks.
• PEP 8 compliant — enforced via ruff (configured in pyproject.toml).
• Lazy imports — heavy converters are only imported when the file type is actually needed.

---

License

MIT