stages: add org.osbuild.copy stage

Copies files obtained via a `source` to the tree. Multiple files or
directories can be copied by specifying multiple entries in `paths`.
If no paths are specified the whole contents of `source` is copied.
The source and the target path for each individual path entry might
optionally be specified via `from` and `to`, respectively; if no
path is given for any of the two, the root `/` is assumed.

Currently only an 'archive' 'source' is supported that in turn uses
the existing 'org.osbuild.files' source to fetch an archive (tarball)
and extracts it to a temporary directory.

stages/org.osbuild.copy

@@ -0,0 +1,138 @@
+#!/usr/bin/python3
+
+import json
+import os
+import sys
+import subprocess
+import tempfile
+
+import osbuild.sources
+
+
+STAGE_DESC = "Copy files from a source to the tree"
+STAGE_INFO = """
+Copies files obtained via a `source` to the tree. Multiple files or
+directories can be copied by specifying multiple entries in `paths`.
+If no paths are specified the whole contents of `source` is copied.
+The source and the target path for each individual path entry might
+optionally be specified via `from` and `to`, respectively; if no
+path is given for any of the two, the root `/` is assumed.
+
+Supported sources are currently:
+
+ o `archive`: Fetch an archive via the org.osbuild.files source and
+   extract it to a temporary directory. Currently only tar archives
+   are supported.
+
+"""
+STAGE_OPTS = """
+"additionalProperties": false,
+"definitions": {
+  "source-archive": {
+    "description": "Fetch an via org.osbuild.files and extract it",
+    "type": "object",
+    "required": ["type", "checksum"],
+    "additionalProperties": false,
+    "properties": {
+      "type": {
+        "type": "string",
+         "description": "The type of the source, here 'archive'",
+         "enum": ["archive"]
+      },
+      "checksum": {
+        "type": "string",
+        "description": "The checksum of the archive to fetch"
+      },
+      "strip-components": {
+        "type": "number",
+        "description": "Strip 'N' leading components from file names on extraction",
+        "default": 1
+      }
+    }
+  }
+},
+"properties": {
+  "source": {
+    "oneOf": [{ "$ref": "#/definitions/source-archive" }]
+  },
+  "paths": {
+    "description": "Array of items to copy",
+    "type": "array",
+    "items": {
+      "type": "object",
+      "additionalProperties": false,
+      "properties": {
+        "from": {
+          "type": "string",
+          "description": "The source path",
+          "default": "/"
+        },
+        "to": {
+          "type": "string",
+          "description": "The destination path",
+          "default": "/"
+        }
+      }
+    }
+  }
+}
+"""
+
+
+def get_archive(options, srcdir, workdir):
+    checksum = options["checksum"]
+    strip = options.get("strip-components", 1)
+    source_path = os.path.join(srcdir, "org.osbuild.files")
+
+    target = os.path.join(workdir, checksum)
+    os.makedirs(target)
+
+    # actually fetch the archive
+    res = osbuild.sources.get("org.osbuild.files", [checksum])
+
+    tar = [
+        "tar",
+        "--strip-components=" + str(strip),
+        "-x",
+        "-C", target,
+        "-f", os.path.join(source_path, checksum)
+    ]
+
+    subprocess.run(tar, check=True)
+
+    return target, res
+
+
+def main(tree, srcdir, options, workdir):
+    source_options = options["source"]
+    source_type = source_options["type"]
+    paths = options.get("paths", [{}])
+
+    if source_type == "archive":
+        source = get_archive
+    else:
+        raise ValueError(f"Unsupported source: {source_type}")
+
+    # Fetch the requested data via the selected source
+    source_path, _ = source(source_options, srcdir, workdir)
+
+    for path in paths:
+        src = path.get("from", "/")
+        dst = path.get("to", "/")
+
+        subprocess.run(["cp", "--reflink=auto", "-a",
+                        f"{source_path}{src}",
+                        f"{tree}{dst}"],
+                       check=True)
+
+
+if __name__ == '__main__':
+    stage_args = json.load(sys.stdin)
+
+    with tempfile.TemporaryDirectory() as _workdir:
+        r = main(stage_args["tree"],
+                 stage_args["sources"],
+                 stage_args["options"],
+                 _workdir)
+
+        sys.exit(r)