BigDiff

BigDiff compares two directory trees and writes a third tree containing only the differences, with inline annotations directly in the files. This repository hosts both official implementations:

• Rust (src/ + Cargo.toml): fast, standalone binary.
• Python (bigdiff/ + bigdiff.py): easy to run/extend; CLI output/messages remain in Portuguese to preserve the original UX.

Choose a version by running its entrypoint; there is no wrapper that forces a selection.

What it does

• Recursively compares Base (A) vs Target (B) and writes results to Output (C).
• New files get a .new suffix; deleted files get .deleted.
• Modified files get .modified with inline diff:
  • Deleted lines are preserved as comments marked DELETED.
  • Added lines are marked NEW.
  • Comment syntax matches the file type (C-like, Python, HTML, etc.).
• Binary or oversized files are copied as .modified with a companion <file>.modified.NOTE.txt explaining why text diff was skipped.
• Supports glob ignores, EOL normalization, and --dry-run to preview the plan.

Repository layout

• src/, Cargo.toml: Rust implementation and dependencies.
• bigdiff/, bigdiff.py: Python package and compatibility wrapper.
• target/: Rust build artifacts (generated by cargo).
• Shared project metadata: README.md, .gitignore, .gitattributes, LICENSE.

Quick start: Rust

Prerequisite: Rust toolchain (cargo + rustc).

Build:

cargo build --release
# binary at target/release/bigdiff

Run:

cargo run -- [OPTIONS] <base_dir> <target_dir> <output_dir>

Notable options:

• -i, --ignore <glob> (repeatable or comma-separated)
• -E, --normalize-eol normalize CRLF/LF before comparing text
• -S, --max-text-size <size> text diff limit (e.g., 5MB)
• --dry-run print the plan without writing files

Example:

cargo run -- ./v1 ./v2 ./diff_out --ignore "node_modules" --ignore "*.log" --normalize-eol

Quick start: Python

Prerequisite: Python 3.8+.

Run:

python -m bigdiff FOLDER1 FOLDER2 FOLDER3 [options]
# compatibility wrapper:
python bigdiff.py FOLDER1 FOLDER2 FOLDER3 [options]

Options (CLI strings are in Portuguese):

• --ignore, -i glob patterns (repeatable or comma-separated)
• --normalize-eol, -E normalize CRLF/LF before comparing text
• --max-text-size, -S text diff limit (default 5MB)
• --dry-run show the summary only

Output structure

• name.ext.new for files only in the target tree.
• name.ext.deleted for files only in the base tree.
• name.ext.modified for files that differ, with inline annotations.
• name.ext.modified.NOTE.txt explains when text diff was skipped.

License

MIT – see LICENSE.