Simplify codefile detection (#5040)

* Deprecate is_string
* Deprecate is_codefile
* Deprecate codefile_type
* Add codefile_class unittest

conda_build/inspect_pkg.py

@@ -24,7 +24,7 @@
     get_package_obj_files,
     get_untracked_obj_files,
 )
-from conda_build.os_utils.liefldd import codefile_type
+from conda_build.os_utils.liefldd import codefile_class, machofile
 from conda_build.os_utils.macho import get_rpaths, human_filetype
 from conda_build.utils import (
     comma_join,
@@ -354,14 +354,16 @@ def inspect_objects(packages, prefix=sys.prefix, groupby="package"):
 
         info = []
         for f in obj_files:
-            f_info = {}
             path = join(prefix, f)
-            filetype = codefile_type(path)
-            if filetype == "machofile":
-                f_info["filetype"] = human_filetype(path, None)
-                f_info["rpath"] = ":".join(get_rpaths(path))
-                f_info["filename"] = f
-                info.append(f_info)
+            codefile = codefile_class(path)
+            if codefile == machofile:
+                info.append(
+                    {
+                        "filetype": human_filetype(path, None),
+                        "rpath": ":".join(get_rpaths(path)),
+                        "filename": f,
+                    }
+                )
 
         output_string += print_object_info(info, groupby)
     if hasattr(output_string, "decode"):

conda_build/os_utils/ldd.py

@@ -8,12 +8,7 @@
 
 from conda_build.conda_interface import linked_data, untracked
 from conda_build.os_utils.macho import otool
-from conda_build.os_utils.pyldd import (
-    codefile_class,
-    inspect_linkages,
-    is_codefile,
-    machofile,
-)
+from conda_build.os_utils.pyldd import codefile_class, inspect_linkages, machofile
 
 LDD_RE = re.compile(r"\s*(.*?)\s*=>\s*(.*?)\s*\(.*\)")
 LDD_NOT_FOUND_RE = re.compile(r"\s*(.*?)\s*=>\s*not found")
@@ -118,7 +113,7 @@ def get_package_obj_files(dist, prefix):
     files = get_package_files(dist, prefix)
     for f in files:
         path = join(prefix, f)
-        if is_codefile(path):
+        if codefile_class(path):
             res.append(f)
 
     return res
@@ -130,7 +125,7 @@ def get_untracked_obj_files(prefix):
     files = untracked(prefix)
     for f in files:
         path = join(prefix, f)
-        if is_codefile(path):
+        if codefile_class(path):
             res.append(f)
 
     return res

conda_build/os_utils/liefldd.py

@@ -1,62 +1,58 @@
 # Copyright (C) 2014 Anaconda, Inc
 # SPDX-License-Identifier: BSD-3-Clause
-try:
-    from collections.abc import Hashable
-except ImportError:
-    from collections.abc import Hashable
+from __future__ import annotations
 
 import hashlib
 import json
 import os
 import struct
 import sys
 import threading
+from collections.abc import Hashable
 from fnmatch import fnmatch
 from functools import partial
+from pathlib import Path
 from subprocess import PIPE, Popen
 
+from ..deprecations import deprecated
 from .external import find_executable
 
 # lief cannot handle files it doesn't know about gracefully
 # TODO :: Remove all use of pyldd
 # Currently we verify the output of each against the other
-from .pyldd import codefile_type as codefile_type_pyldd
+from .pyldd import DLLfile, EXEfile, elffile, machofile
+from .pyldd import codefile_type as _codefile_type
 from .pyldd import inspect_linkages as inspect_linkages_pyldd
 
-codefile_type = codefile_type_pyldd
-have_lief = False
 try:
     import lief
 
     lief.logging.disable()
     have_lief = True
-except:
-    pass
+except ImportError:
+    have_lief = False
 
 
+@deprecated("3.28.0", "4.0.0", addendum="Use `isinstance(value, str)` instead.")
 def is_string(s):
-    try:
-        return isinstance(s, basestring)
-    except NameError:
-        return isinstance(s, str)
+    return isinstance(s, str)
 
 
 # Some functions can operate on either file names
 # or an already loaded binary. Generally speaking
 # these are to be avoided, or if not avoided they
 # should be passed a binary when possible as that
 # will prevent having to parse it multiple times.
-def ensure_binary(file):
-    if not is_string(file):
+def ensure_binary(file: str | os.PathLike | Path | lief.Binary) -> lief.Binary | None:
+    if isinstance(file, lief.Binary):
         return file
-    else:
-        try:
-            if not os.path.exists(file):
-                return []
-            return lief.parse(file)
-        except:
-            print(f"WARNING: liefldd: failed to ensure_binary({file})")
-    return None
+    elif not Path(file).exists():
+        return None
+    try:
+        return lief.parse(str(file))
+    except BaseException:
+        print(f"WARNING: liefldd: failed to ensure_binary({file})")
+        return None
 
 
 def nm(filename):
@@ -77,25 +73,57 @@ def nm(filename):
         print("No symbols found")
 
 
-def codefile_type_liefldd(file, skip_symlinks=True):
-    binary = ensure_binary(file)
-    result = None
-    if binary:
-        if binary.format == lief.EXE_FORMATS.PE:
-            if lief.PE.DLL_CHARACTERISTICS:
-                if binary.header.characteristics & lief.PE.HEADER_CHARACTERISTICS.DLL:
-                    result = "DLLfile"
-                else:
-                    result = "EXEfile"
+if have_lief:
+
+    def codefile_class(
+        path: str | os.PathLike | Path,
+        skip_symlinks: bool = False,
+    ) -> type[DLLfile | EXEfile | machofile | elffile] | None:
+        # same signature as conda.os_utils.pyldd.codefile_class
+        if not (binary := ensure_binary(path)):
+            return None
+        elif (
+            binary.format == lief.EXE_FORMATS.PE
+            and lief.PE.HEADER_CHARACTERISTICS.DLL in binary.header.characteristics_list
+        ):
+            return DLLfile
+        elif binary.format == lief.EXE_FORMATS.PE:
+            return EXEfile
         elif binary.format == lief.EXE_FORMATS.MACHO:
-            result = "machofile"
+            return machofile
         elif binary.format == lief.EXE_FORMATS.ELF:
-            result = "elffile"
-    return result
-
+            return elffile
+        else:
+            return None
 
-if have_lief:
-    codefile_type = codefile_type_liefldd
+else:
+    from .pyldd import codefile_class
+
+
+@deprecated(
+    "3.28.0",
+    "4.0.0",
+    addendum="Use `conda_build.os_utils.liefldd.codefile_class` instead.",
+)
+def codefile_type_liefldd(*args, **kwargs) -> str | None:
+    codefile = codefile_class(*args, **kwargs)
+    return codefile.__name__ if codefile else None
+
+
+deprecated.constant(
+    "3.28.0",
+    "4.0.0",
+    "codefile_type_pyldd",
+    _codefile_type,
+    addendum="Use `conda_build.os_utils.pyldd.codefile_class` instead.",
+)
+deprecated.constant(
+    "3.28.0",
+    "4.0.0",
+    "codefile_type",
+    _codefile_type,
+    addendum="Use `conda_build.os_utils.liefldd.codefile_class` instead.",
+)
 
 
 def _trim_sysroot(sysroot):
@@ -111,7 +139,9 @@ def get_libraries(file):
         if binary.format == lief.EXE_FORMATS.PE:
             result = binary.libraries
         else:
-            result = [lib if is_string(lib) else lib.name for lib in binary.libraries]
+            result = [
+                lib if isinstance(lib, str) else lib.name for lib in binary.libraries
+            ]
             # LIEF returns LC_ID_DYLIB name @rpath/libbz2.dylib in binary.libraries. Strip that.
             binary_name = None
             if binary.format == lief.EXE_FORMATS.MACHO:
@@ -505,7 +535,7 @@ def inspect_linkages_lief(
                     while tmp_filename:
                         if (
                             not parent_exe_dirname
-                            and codefile_type(tmp_filename) == "EXEfile"
+                            and codefile_class(tmp_filename) == EXEfile
                         ):
                             parent_exe_dirname = os.path.dirname(tmp_filename)
                         tmp_filename = parents_by_filename[tmp_filename]
@@ -595,7 +625,7 @@ def get_linkages(
     result_pyldd = []
     debug = False
     if not have_lief or debug:
-        if codefile_type(filename) not in ("DLLfile", "EXEfile"):
+        if codefile_class(filename) not in (DLLfile, EXEfile):
             result_pyldd = inspect_linkages_pyldd(
                 filename,
                 resolve_filenames=resolve_filenames,
@@ -607,7 +637,7 @@ def get_linkages(
                 return result_pyldd
         else:
             print(
-                f"WARNING: failed to get_linkages, codefile_type('{filename}')={codefile_type(filename)}"
+                f"WARNING: failed to get_linkages, codefile_class('{filename}')={codefile_class(filename)}"
             )
             return {}
     result_lief = inspect_linkages_lief(

conda_build/os_utils/pyldd.py

@@ -1,15 +1,20 @@
 # Copyright (C) 2014 Anaconda, Inc
 # SPDX-License-Identifier: BSD-3-Clause
+from __future__ import annotations
+
 import argparse
 import glob
 import logging
 import os
 import re
 import struct
 import sys
+from pathlib import Path
 
 from conda_build.utils import ensure_list, get_logger
 
+from ..deprecations import deprecated
+
 logging.basicConfig(level=logging.INFO)
 
 
@@ -1028,46 +1033,60 @@ def codefile(file, arch="any", initial_rpaths_transitive=[]):
         return inscrutablefile(file, list(initial_rpaths_transitive))
 
 
-def codefile_class(filename, skip_symlinks=False):
-    if os.path.islink(filename):
-        if skip_symlinks:
-            return None
-        else:
-            filename = os.path.realpath(filename)
-    if os.path.isdir(filename):
+def codefile_class(
+    path: str | os.PathLike | Path,
+    skip_symlinks: bool = False,
+) -> type[DLLfile | EXEfile | machofile | elffile] | None:
+    # same signature as conda.os_utils.liefldd.codefile_class
+    path = Path(path)
+    if skip_symlinks and path.is_symlink():
         return None
-    if filename.endswith((".dll", ".pyd")):
+    path = path.resolve()
+
+    def _get_magic_bit(path: Path) -> bytes:
+        with path.open("rb") as handle:
+            bit = handle.read(4)
+        return struct.unpack(BIG_ENDIAN + "L", bit)[0]
+
+    if path.is_dir():
+        return None
+    elif path.suffix.lower() in (".dll", ".pyd"):
         return DLLfile
-    if filename.endswith(".exe"):
+    elif path.suffix.lower() == ".exe":
         return EXEfile
-    # Java .class files share 0xCAFEBABE with Mach-O FAT_MAGIC.
-    if filename.endswith(".class"):
+    elif path.suffix.lower() == ".class":
+        # Java .class files share 0xCAFEBABE with Mach-O FAT_MAGIC.
         return None
-    if not os.path.exists(filename) or os.path.getsize(filename) < 4:
+    elif not path.exists() or path.stat().st_size < 4:
+        return None
+    elif (magic := _get_magic_bit(path)) == ELF_HDR:
+        return elffile
+    elif magic in (FAT_MAGIC, MH_MAGIC, MH_CIGAM, MH_CIGAM_64):
+        return machofile
+    else:
         return None
-    with open(filename, "rb") as file:
-        (magic,) = struct.unpack(BIG_ENDIAN + "L", file.read(4))
-        file.seek(0)
-        if magic in (FAT_MAGIC, MH_MAGIC, MH_CIGAM, MH_CIGAM_64):
-            return machofile
-        elif magic == ELF_HDR:
-            return elffile
-    return None
 
 
-def is_codefile(filename, skip_symlinks=True):
-    klass = codefile_class(filename, skip_symlinks=skip_symlinks)
-    if not klass:
-        return False
-    return True
+@deprecated(
+    "3.28.0",
+    "4.0.0",
+    addendum="Use `conda_build.os_utils.pyldd.codefile_class` instead.",
+)
+def is_codefile(path: str | os.PathLike | Path, skip_symlinks: bool = True) -> bool:
+    return bool(codefile_class(path, skip_symlinks=skip_symlinks))
 
 
-def codefile_type(filename, skip_symlinks=True):
-    "Returns None, 'machofile' or 'elffile'"
-    klass = codefile_class(filename, skip_symlinks=skip_symlinks)
-    if not klass:
-        return None
-    return klass.__name__
+@deprecated(
+    "3.28.0",
+    "4.0.0",
+    addendum="Use `conda_build.os_utils.pyldd.codefile_class` instead.",
+)
+def codefile_type(
+    path: str | os.PathLike | Path,
+    skip_symlinks: bool = True,
+) -> str | None:
+    codefile = codefile_class(path, skip_symlinks=skip_symlinks)
+    return codefile.__name__ if codefile else None
 
 
 def _trim_sysroot(sysroot):