wip heritrix package

2025-08-10 13:33:15 -07:00
parent 8f1dfd462d
commit c477949143
1 changed files with 31 additions and 0 deletions
--- a/packages/heritrix-bin/package.nix
+++ b/packages/heritrix-bin/package.nix
@@ -0,0 +1,31 @@
+{
+  lib,
+  stdenv,
+  fetchurl,
+}:
+let
+  # list from https://github.com/hsjobeki/nixpkgs/blob/f0efec9cacfa9aef98a3c70fda2753b9825e262f/pkgs/build-support/fetchmavenartifact/default.nix#L9
+  defaultRepos = [
+    "https://repo1.maven.org/maven2"
+    "https://oss.sonatype.org/content/repositories/releases"
+    "https://oss.sonatype.org/content/repositories/public"
+    "https://repo.typesafe.com/typesafe/releases"
+  ];
+in
+stdenv.mkDerivation (finalAttrs: {
+  pname = "heritrix-bin";
+  version = "3.10.1";
+
+  # https://repo1.maven.org/maven2/org/archive/heritrix/heritrix/3.10.1/heritrix-3.10.1-dist.tar.gz
+  src = fetchurl rec {
+    name = "heritrix-${finalAttrs.version}-dist.tar.gz";
+    urls = map (repo: "${repo}/org/archive/heritrix/heritrix/${finalAttrs.version}/${name}") defaultRepos;
+    hash = "sha256-4qf2+dgmRgyD58cQB0jiT+y4XZ3+GjRjhADk+bewt7Q=";
+  };
+
+  meta = {
+    description = "open-source, extensible, archival-quality web crawler project by Internet Archive";
+    license = lib.licenses.asl20;
+    sourceProvenance = with lib.sourceTypes; [ binaryBytecode binaryNativeCode ];
+  };
+})