.gitattributes

# ============================================================
# .gitattributes (Keep files consistent across operating systems)
# ============================================================

# REQ.UNIVERSAL: All professional GitHub project repositories MUST include .gitattributes.
# WHY: Ensure consistent line endings, diff behavior, and file classification
# across Windows, macOS, and Linux environments.
# ALT: Repository may omit .gitattributes ONLY if equivalent normalization is
# enforced reliably by tooling and CI (rare and fragile).
# CUSTOM: Update file-type rules only when introducing new languages,
# binary artifacts, or documentation formats.
# NOTE: Rules are ordered by impact and generality, not alphabetically.
# Git attributes are documented at https://git-scm.com/docs/gitattributes

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

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

# WHY-SECTION: Explicit EOL rules avoid platform-specific diffs and tool failures.

# === Scripts ===
# Python and shell scripts must always use LF (CI, Linux, macOS, containers)
*.py  text eol=lf
*.sh  text eol=lf

# PowerShell scripts follow Windows convention
*.ps1 text eol=crlf

# === Docs and configs ===
# Cross-platform; tooling expects LF
*.md   text eol=lf
*.yaml text eol=lf
*.yml  text eol=lf
*.json text eol=lf
*.toml text eol=lf

# === GitHub Linguist ===
# Improve language statistics and UI without affecting builds
docs/** linguist-documentation

# Jupyter notebooks: custom diff/merge drivers (if configured)
*.ipynb diff=jupyternotebook
*.ipynb merge=jupyternotebook



# === Source Code ===

# 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.
*.bib   text eol=lf
*.cls   text eol=lf
*.md    text eol=lf
*.sty   text eol=lf
*.tex   text eol=lf


# === Configuration and Data (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


# === GitHub Metadata ===

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


# === 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 / 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


# === Proof Assistants ===

# WHY: Lean 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

# === Lean Build System ===

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