Implement readlink -f with libc's realpath() (#147)

This is how busybox does it.  coreutils readlink implements its own algorithm in hundreds of lines of C.

Python's os.path.realpath() is too lenient.


Co-authored-by: Caroline Lin <caroline-lin@users.noreply.github.com>
Co-authored-by: Andy Chu <andychu@users.noreply.github.com>

bin/readlink

@@ -0,0 +1 @@
oil.py

gold/readlink.sh

@@ -8,11 +8,25 @@ set -o pipefail
#set -o errexit

dir-does-not-exist() {
  readlink -f _tmp/gold-bin/readlink
  echo $?

  readlink -f libc.so
  echo $?

  readlink -f /nonexistent
  echo $?

  readlink -f /nonexistent/foo
  echo $?
}

compare() {
  PATH="_tmp/gold-bin:$PATH" $0 dir-does-not-exist > _tmp/busybox-readlink.txt

  PATH="bin/:$PATH" $0 dir-does-not-exist > _tmp/osh-readlink.txt

  diff -u _tmp/busybox-readlink.txt _tmp/osh-readlink.txt
}

"$@"

native/libc.c

@@ -4,6 +4,8 @@

#include <stdarg.h>  // va_list, etc.
#include <stdio.h>  // printf
#include <limits.h>
#include <stdlib.h>

#include <fnmatch.h>
#include <glob.h>
@@ -26,6 +28,24 @@ static void debug(const char* fmt, ...) {
#endif
}

static PyObject *
func_realpath(PyObject *self, PyObject *args) {
  const char *symlink;

  if (!PyArg_ParseTuple(args, "s", &symlink)) {
    return NULL;
  }
  char target[PATH_MAX + 1];
  char *status = realpath(symlink, target);

  if (status== NULL) {
      debug("error occurred while resolving symlink");
      return PyLong_FromLong(-1);
  } else {
      return PyString_FromString(target);
  }
}

static PyObject *
func_fnmatch(PyObject *self, PyObject *args) {
  const char *pattern;
@@ -309,6 +329,8 @@ func_regex_first_group_match(PyObject *self, PyObject *args) {
}

static PyMethodDef methods[] = {
  {"realpath", func_realpath, METH_VARARGS,
   "Return canonical path of a symlink."},
  {"fnmatch", func_fnmatch, METH_VARARGS,
   "Return whether a string matches a pattern."},
  // Python's glob doesn't have char classes

native/libc_test.py

@@ -8,7 +8,6 @@
"""
libc_test.py: Tests for libc.py
"""

import unittest

import libc  # module under test
@@ -94,6 +93,21 @@ def testRegexFirstGroupMatch(self):
        (8, 10),
        libc.regex_first_group_match('(X.)', s, 6))

  def testRealpathFailOnNonexistentDirectory(self):
	# This behaviour is actually inconsistent with GNU readlink,
	# but matches behaviour of busybox readlink
	# (https://github.com/jgunthorpe/busybox)
	self.assertEqual(
		-1,
		libc.realpath('_tmp/nonexistent')
	)

	# Consistent with GNU
	self.assertEqual(
		-1,
		libc.realpath('_tmp/nonexistent/supernonexistent')
	)


if __name__ == '__main__':
  unittest.main()

tools/readlink.py

@@ -4,7 +4,7 @@
readlink.py
"""

import os
import libc
from core import args, util

SPEC = args.BuiltinFlags()
@@ -17,6 +17,8 @@ def main(argv):
    util.error("-f must be passed")
    return 1
  for arg in argv[i:]:
    res = os.path.realpath(arg)
    res = libc.realpath(arg)
    if res == -1:
        return 1
    print(res)
  return 0