patdiff
File diff using the patience diff algorithm: a MoonBit port of Jane Street's patdiff
diff
patience
ansi
Author
Version
0.1.0
License
MIT
Repository
Last updated
4 hours ago
Downloads
3
#patdiff.mbt
A MoonBit port of patdiff, Jane Street's
file diff using the patience diff algorithm. It highlights the words that changed
within lines, and renders to the terminal (ANSI or plain ASCII), to HTML, and side
by side.
The port follows upstream at v0.18~preview.130.106+341, together with the
patience_diff library it depends on.
#Command line
The native build gives a patdiff executable, with upstream's flags:
moon build --release_build/native/release/build/cmd/patdiff/patdiff.exe old.txt new.txtAs upstream it compares two files or two directories, or reads diff-like text from
stdin when given no files. Flags can be abbreviated to any unique prefix. ~/.patdiff
is read as a configuration file if it exists; -make-config FILE writes a sample one.
The exit code is 0 for no differences and 1 for differences or an error.
Useful flags: -html, -ascii, -side-by-side wrap, -find-moves, -unrefined,
-keep-whitespace, -context NUM, -float-tolerance 1%, -location-style omake.
See -help for all of them.
scripts/patdiff-git-wrapper is upstream's wrapper for using patdiff as git's
external diff.
#Library
The kernel package is pure MoonBit, so it isn't tied to the native backend:
let diff : Bytes = @kernel.patdiff(
prev=@kernel.DiffInput::of_string("old", "Foo bar buzz"),
next=@kernel.DiffInput::of_string("new", "Foo buzz"),
output=Html,
)| Package | Contents |
|---|---|
| patience_diff | The diff algorithm, generic over the element type |
| ansi_text | Parsing and manipulating text with ANSI escape codes |
| kernel | Refinement, moves, rendering and configuration; no I/O |
| sexp | S-expressions, for configuration files |
| sys | Native bindings for files and the terminal |
| lib | Comparing files and directories, loading configuration; native only |
| cmd/patdiff, cmd/ansi_text | The executables; native only |
All text is Bytes, because file contents needn't be valid UTF-8, and OCaml's
string is a byte sequence too. DiffInput::of_string encodes a String as UTF-8,
and @utf8.decode_lossy from moonbitlang/core/encoding/utf8 decodes the result.
#Differences from upstream
- -include and -exclude use MoonBit's @string.Regex rather than OCaml's Re.Pcre, so the regexp syntax isn't identical.
- Very large replaced regions: when a single Replace has hundreds of lines, upstream breaks it up around rare words, and breaks ties between equally rare words by hash table order. This port uses first appearance, so a tie may be broken differently.
- Terminal width is read from the controlling terminal on every Unix. Upstream only does so on Linux, using 121 columns elsewhere.
- Command line errors and -help have the same content as Core's Command, but not exactly the same layout.
- -double-check compares the files in-process rather than running cmp.
- SGR parameters above 2^31 - 1 are treated as missing, because they don't fit in an Int.
- Upstream's Patience_diff.merge isn't used by patdiff, and isn't ported.
#Testing
moon testmoon build --release && python3 test/cli/run_cram.pymoon test runs upstream's expect tests, ported to MoonBit, on any backend. test/cli
holds upstream's command line tests as cram files; they need bash and git.
Install
Download zipFile diff using the patience diff algorithm: a MoonBit port of Jane Street's patdiff
diffpatienceansi
MIT
Updated 4 hours ago
3 downloads