nix-gitignore: Optimise performance

There are a few operations in this library that naively runs on every
iteration while they could be cached.

For a simple test repository with a small number of files and ~1000
gitignore patterns this brings memory usage down from ~233M to ~157M
and wall time from 2.6s down to 0.78s.

This should scale similarly with the number of files in a repository.
This commit is contained in:
adisbladis 2020-12-07 13:42:59 +01:00
parent a817ce3ada
commit ef3ed45c12
No known key found for this signature in database
GPG Key ID: 110BFAD44C6249B7

View File

@ -20,14 +20,13 @@ let
in rec {
# [["good/relative/source/file" true] ["bad.tmpfile" false]] -> root -> path
filterPattern = patterns: root:
(name: _type:
let
relPath = lib.removePrefix ((toString root) + "/") name;
matches = pair: (match (head pair) relPath) != null;
matched = map (pair: [(matches pair) (last pair)]) patterns;
in
last (last ([[true true]] ++ (filter head matched)))
);
let
filters = map (pair: relPath: if match (head pair) relPath == null then true else last pair) patterns;
in
name: _type:
let
relPath = lib.removePrefix ((toString root) + "/") name;
in foldl' (acc: f: if acc == true then f relPath else acc) true filters;
# string -> [[regex bool]]
gitignoreToPatterns = gitignore:
@ -91,7 +90,9 @@ in rec {
(filter (l: !isList l && !isComment l)
(split "\n" gitignore));
gitignoreFilter = ign: root: filterPattern (gitignoreToPatterns ign) root;
gitignoreFilter = ign: let
patterns = gitignoreToPatterns ign;
in root: filterPattern patterns root;
# string|[string|file] (→ [string|file] → [string]) -> string
gitignoreCompileIgnore = file_str_patterns: root:
@ -100,9 +101,10 @@ in rec {
str_patterns = map (onPath readFile) (lib.toList file_str_patterns);
in concatStringsSep "\n" str_patterns;
gitignoreFilterPure = filter: patterns: root: name: type:
gitignoreFilter (gitignoreCompileIgnore patterns root) root name type
&& filter name type;
gitignoreFilterPure = filter: patterns: root: let
compiledFilter = gitignoreCompileIgnore patterns root;
filterFn = gitignoreFilter compiledFilter;
in name: type: filterFn root name type && filter name type;
# This is a very hacky way of programming this!
# A better way would be to reuse existing filtering by making multiple gitignore functions per each root.