Repair ELF executables in the "scripts" directory

.gitignore

@@ -62,4 +62,5 @@ target/
 # Generated by test script
 *.zip
 wheelhoust-*
-tests/testpackage/testpackage/testprogram
+tests/integration/testpackage/testpackage/testprogram
+tests/integration/testpackage/testpackage/testprogram_nodeps

src/auditwheel/repair.py

@@ -66,7 +66,7 @@ def repair_wheel(
         if not exists(dest_dir):
             os.mkdir(dest_dir)
 
-        # here, fn is a path to a python extension library in
+        # here, fn is a path to an ELF file (lib or executable) in
         # the wheel, and v['libs'] contains its required libs
         for fn, v in external_refs_by_fn.items():
             ext_libs: dict[str, str] = v[abis[0]]["libs"]
@@ -92,6 +92,9 @@ def repair_wheel(
                 patcher.replace_needed(fn, *replacements)
 
             if len(ext_libs) > 0:
+                if _path_is_script(fn):
+                    fn = _replace_elf_script_with_shim(match.group("name"), fn)
+
                 new_rpath = os.path.relpath(dest_dir, os.path.dirname(fn))
                 new_rpath = os.path.join("$ORIGIN", new_rpath)
                 append_rpath_within_wheel(fn, new_rpath, ctx.name, patcher)
@@ -232,3 +235,58 @@ def _resolve_rpath_tokens(rpath: str, lib_base_dir: str) -> str:
         rpath = rpath.replace(f"${token}", target)  # $TOKEN
         rpath = rpath.replace(f"${{{token}}}", target)  # ${TOKEN}
     return rpath
+
+
+def _path_is_script(path: str) -> bool:
+    # Looks something like "uWSGI-2.0.21.data/scripts/uwsgi"
+    components = path.split("/")
+    return (
+        len(components) == 3
+        and components[0].endswith(".data")
+        and components[1] == "scripts"
+    )
+
+
+def _replace_elf_script_with_shim(package_name: str, orig_path: str) -> str:
+    """Move an ELF script and replace it with a shim.
+
+    We can't directly rewrite the RPATH of ELF executables in the "scripts"
+    directory since scripts aren't installed to a consistent relative path to
+    platlib files.
+
+    Instead, we move the executable into a special directory in platlib and put
+    a shim script in its place which execs the real executable.
+
+    More context: https://github.com/pypa/auditwheel/issues/340
+
+    Returns the new path of the moved executable.
+    """
+    scripts_dir = f"{package_name}.scripts"
+    os.makedirs(scripts_dir, exist_ok=True)
+
+    new_path = os.path.join(scripts_dir, os.path.basename(orig_path))
+    os.rename(orig_path, new_path)
+
+    with open(orig_path, "w") as f:
+        f.write(_script_shim(new_path))
+    os.chmod(orig_path, os.stat(new_path).st_mode)
+
+    return new_path
+
+
+def _script_shim(binary_path: str) -> str:
+    return """\
+#!python
+import os
+import sys
+import sysconfig
+
+
+if __name__ == "__main__":
+    os.execv(
+        os.path.join(sysconfig.get_path("platlib"), {binary_path!r}),
+        sys.argv,
+    )
+""".format(
+        binary_path=binary_path,
+    )

tests/integration/test_manylinux.py

@@ -175,7 +175,7 @@ def tmp_docker_image(base, commands, setup_env={}):
 
     logger.info("Made image %s based on %s", image.short_id, base)
     try:
-        yield image.short_id
+        yield image.id
     finally:
         client = image.client
         client.images.remove(image.id)
@@ -404,6 +404,33 @@ def test_build_wheel_with_binary_executable(
         )
         assert output.strip() == "2.25"
 
+        # Both testprogram and testprogram_nodeps square a number, but:
+        #   * testprogram links against libgsl and had to have its RPATH
+        #     rewritten.
+        #   * testprogram_nodeps links against no shared libraries and wasn't
+        #     rewritten.
+        #
+        # Both executables should work when called from the installed bin directory.
+        assert docker_exec(docker_python, ["/usr/local/bin/testprogram", "4"]) == "16\n"
+        assert (
+            docker_exec(docker_python, ["/usr/local/bin/testprogram_nodeps", "4"])
+            == "16\n"
+        )
+
+        # testprogram should be a Python shim since we had to rewrite its RPATH.
+        assert (
+            docker_exec(docker_python, ["head", "-n1", "/usr/local/bin/testprogram"])
+            == "#!/usr/local/bin/python\n"
+        )
+
+        # testprogram_nodeps should be the unmodified ELF binary.
+        assert (
+            docker_exec(
+                docker_python, ["head", "-c4", "/usr/local/bin/testprogram_nodeps"]
+            )
+            == "\x7fELF"
+        )
+
     def test_build_repair_pure_wheel(self, any_manylinux_container, io_folder):
         policy, tag, manylinux_ctr = any_manylinux_container
 

tests/integration/testpackage/setup.py

@@ -4,12 +4,32 @@
 
 from setuptools import setup
 
-cmd = "gcc testpackage/testprogram.c -lgsl -lgslcblas -o testpackage/testprogram"
-subprocess.check_call(cmd.split())
+subprocess.check_call(
+    (
+        "gcc",
+        "testpackage/testprogram.c",
+        "-lgsl",
+        "-lgslcblas",
+        "-o",
+        "testpackage/testprogram",
+    )
+)
+subprocess.check_call(
+    ("gcc", "testpackage/testprogram_nodeps.c", "-o", "testpackage/testprogram_nodeps")
+)
 
 setup(
     name="testpackage",
     version="0.0.1",
     packages=["testpackage"],
-    package_data={"testpackage": ["testprogram"]},
+    package_data={"testpackage": ["testprogram", "testprogram_nodeps"]},
+    # This places these files at a path like
+    # "testpackage-0.0.1.data/scripts/testprogram", which is needed to test
+    # rewriting ELF binaries installed into the scripts directory.
+    #
+    # Note that using scripts=[] doesn't work here since setuptools expects the
+    # scripts to be text and tries to decode them using UTF-8.
+    data_files=[
+        ("../scripts", ["testpackage/testprogram", "testpackage/testprogram_nodeps"])
+    ],
 )

tests/integration/testpackage/testpackage/testprogram_nodeps.c

@@ -0,0 +1,19 @@
+/* A simple example program to square a number using no shared libraries. */
+
+#include <stdio.h>
+#include <stdlib.h>
+
+int main(int argc, char **argv)
+{
+    int x;
+
+    if (argc != 2)
+    {
+        fputs("Expected exactly one command line argument\n", stderr);
+        return EXIT_FAILURE;
+    }
+
+    x = atoi(argv[1]);
+    printf("%d\n", x*x);
+    return EXIT_SUCCESS;
+}