.gitattributes

# REQ.UNIVERSAL: All professional GitHub project repositories MUST include .gitattributes.
# WHY-FILE: Ensure consistent line endings, diff behavior, and repository metadata
# across Windows, macOS, and Linux.
# ALT: Repository may omit .gitattributes ONLY if equivalent normalization is
# enforced by tooling and CI (rare and fragile).
# CUSTOM: Update file-type rules only when introducing new languages or artifact types.

# === Core defaults (always apply) ===

# WHY: Auto-detect text files and normalize line endings to avoid cross-platform drift.
* text=auto


# === Programming languages and scripts ===

# WHY: Python and shell scripts must use LF for CI/CD, Linux environments, and containers.
*.py    text eol=lf
*.sh    text eol=lf

# WHY: PowerShell convention on Windows uses CRLF.
*.ps1   text eol=crlf


# === Markup and documentation ===

# WHY: Documentation and markup files use LF; standard for cross-platform tooling.
*.md    text eol=lf
*.tex   text eol=lf
*.sty   text eol=lf
*.cls   text eol=lf
*.bib   text eol=lf


# === Configuration and structured text ===

# WHY: Configuration and structured text formats use LF for stable diffs.
*.json     text eol=lf
*.jsonc   text eol=lf
*.jsonl   text eol=lf
*.ndjson  text eol=lf
*.toml    text eol=lf
*.yaml    text eol=lf
*.yml     text eol=lf


# === Proof assistants and formal languages ===

# WHY: Lean source files must use LF for cross-platform consistency and CI.
*.lean   text eol=lf

# WHY: Lean build artifacts are binary; prevent normalization and meaningless diffs.
*.olean  binary
*.ilean  binary
*.trace  binary

# WHY: Coq source uses LF; compiled objects are binary.
*.v      text eol=lf
*.vo     binary
*.vok    binary
*.vos    binary
*.glob   binary

# WHY: Lake build directory should be excluded, but if tracked, treat as binary.
.lake/**  binary


# === Notebooks ===

# WHY: Jupyter notebooks require specialized diff and merge handling.
*.ipynb   diff=jupyternotebook
*.ipynb   merge=jupyternotebook


# === Databases (binary) ===

# WHY: Database files are binary; prevent text normalization and meaningless diffs.
*.db       binary
*.duckdb   binary
*.sqlite   binary
*.sqlite3  binary


# === Columnar and analytical data (binary) ===

# WHY: Columnar and analytical data formats are binary; diffs are not meaningful.
*.arrow    binary
*.avro     binary
*.feather  binary
*.orc      binary
*.parquet  binary


# === Office, BI, PDFs, and compressed artifacts (binary) ===

# WHY: Office documents, BI files, PDFs, and compressed artifacts are binary.
*.7z     binary
*.bz2    binary
*.docx   binary
*.gz     binary
*.pbix   binary
*.pbit   binary
*.pdf    binary
*.pptx   binary
*.rar    binary
*.tar    binary
*.tgz    binary
*.xls    binary
*.xlsm   binary
*.xlsx   binary
*.xz     binary
*.zip    binary


# === GitHub metadata and UI ===

# WHY: Exclude documentation and tests from GitHub language statistics.
docs/**    linguist-documentation
tests/**   linguist-documentation