Consolidate all the adhoc file magic tests

This should also speed up namcap a little, since it replaces full
extraction and tempfiles with a small in-place read().

Signed-off-by: Kyle Keen <keenerd@gmail.com>

8 files changed, 32 insertions, 30 deletions

diff --git a/Namcap/depends.py b/Namcap/depends.py
index 3c115bf..af11c88 100644
--- a/Namcap/depends.py
+++ b/Namcap/depends.py
@@ -24,7 +24,6 @@
 import re, os, os.path
 import subprocess
 import tempfile
-from Namcap.util import is_elf, script_type
 from Namcap.ruleclass import *
 import Namcap.tags
 from Namcap import package
diff --git a/Namcap/rules/anyelf.py b/Namcap/rules/anyelf.py
index 90172a3..1e497e4 100644
--- a/Namcap/rules/anyelf.py
+++ b/Namcap/rules/anyelf.py
@@ -22,7 +22,7 @@ Check for ELF files to see if a package should be 'any' architecture
 """
 
 import os, re
-from Namcap.util import is_elf, clean_filename
+from Namcap.util import is_elf, is_static, clean_filename
 from Namcap.ruleclass import *
 
 class package(TarballRule):
@@ -38,8 +38,8 @@ class package(TarballRule):
 			if not entry.isfile():
 				continue
 			f = tar.extractfile(entry)
-			# Archive files are considered as ELF (FS#24854)
-			if f.read(4) in (b"\x7fELF", b"!<ar"):
+			# Ar files (static libs) are also architecture specific (FS#24854)
+			if is_elf(f) or is_static(f):
 				found_elffiles.append(entry.name)
 			f.close()
 
diff --git a/Namcap/rules/elffiles.py b/Namcap/rules/elffiles.py
index f8f16ac..68167bd 100644
--- a/Namcap/rules/elffiles.py
+++ b/Namcap/rules/elffiles.py
@@ -51,7 +51,7 @@ class ELFPaths(TarballRule):
 				continue
 			# is it an ELF file ?
 			f = tar.extractfile(entry)
-			if f.read(4) == b"\x7fELF":
+			if is_elf(f):
 				invalid_elffiles.append(entry.name)
 
 		self.errors = [("elffile-not-in-allowed-dirs %s", i)
@@ -62,13 +62,12 @@ def _test_elf_and_extract(tar, entry):
 	if not entry.isfile():
 		return
 	f = tar.extractfile(entry)
-	magic = f.read(4)
-	if magic != b"\x7fELF":
+	if not is_elf(f):
 		return
 
 	# read the rest of file
 	tmp = tempfile.NamedTemporaryFile(delete=False)
-	tmp.write(magic + f.read())
+	tmp.write(f.read())
 	tmp.close()
 	return tmp.name
 
diff --git a/Namcap/rules/rpath.py b/Namcap/rules/rpath.py
index a2d1193..f8e918f 100644
--- a/Namcap/rules/rpath.py
+++ b/Namcap/rules/rpath.py
@@ -58,11 +58,11 @@ class package(TarballRule):
 				continue
 
 			# is it an ELF file ?
+			if not is_elf(tar.extractfile(entry)):
+				continue # not an ELF file
 			f = tar.extractfile(entry)
 			elf = f.read()
 			f.close()
-			if elf[:4] != b"\x7fELF":
-				continue # not an ELF file
 
 			# write it to a temporary file
 			f = tempfile.NamedTemporaryFile(delete = False)
diff --git a/Namcap/rules/shebangdepends.py b/Namcap/rules/shebangdepends.py
index bac153f..088d888 100644
--- a/Namcap/rules/shebangdepends.py
+++ b/Namcap/rules/shebangdepends.py
@@ -27,7 +27,7 @@ import tempfile
 import shutil
 import pyalpm
 import Namcap.package
-from Namcap.util import script_type
+from Namcap.util import is_script, script_type
 from Namcap.ruleclass import *
 
 def scanshebangs(fileobj, filename, scripts):
@@ -39,12 +39,11 @@ def scanshebangs(fileobj, filename, scripts):
 	"""
 
 	# test magic bytes
-	magic = fileobj.read(2)
-	if magic != b"#!":
+	if not is_script(fileobj):
 		return
 	# read the rest of file
 	tmp = tempfile.NamedTemporaryFile(delete=False)
-	tmp.write(magic + fileobj.read())
+	tmp.write(fileobj.read())
 	tmp.close()
 
 	try:
diff --git a/Namcap/rules/sodepends.py b/Namcap/rules/sodepends.py
index 6832ae0..bb17cb1 100644
--- a/Namcap/rules/sodepends.py
+++ b/Namcap/rules/sodepends.py
@@ -27,6 +27,7 @@ import os
 import subprocess
 import Namcap.package
 from Namcap.ruleclass import *
+from Namcap.util import is_elf
 
 from elftools.elf.enums import ENUM_D_TAG
 from elftools.elf.elffile import ELFFile
@@ -43,12 +44,9 @@ def scanlibs(fileobj, filename):
 	returns: a dictionary { library => set(ELF files using that library) }
 	"""
 
-	# test magic bytes
-	magic = fileobj.read(4)
-	if magic[:4] != b"\x7fELF":
+	if not is_elf(fileobj):
 		return {}
 
-	fileobj.seek(0)
 	elffile = ELFFile(fileobj)
 	sharedlibs = defaultdict(set)
 	for section in elffile.iter_sections():
diff --git a/Namcap/rules/unusedsodepends.py b/Namcap/rules/unusedsodepends.py
index 82bfcab..fe2045c 100644
--- a/Namcap/rules/unusedsodepends.py
+++ b/Namcap/rules/unusedsodepends.py
@@ -54,11 +54,11 @@ class package(TarballRule):
 				continue
 
 			# is it an ELF file ?
+			if not is_elf(tar.extractfile(entry)):
+				continue # not an ELF file
 			f = tar.extractfile(entry)
 			elf = f.read()
 			f.close()
-			if elf[:4] != b"\x7fELF":
-				continue # not an ELF file
 
 			# write it to a temporary file
 			f = tempfile.NamedTemporaryFile(delete = False)
diff --git a/Namcap/util.py b/Namcap/util.py
index 1b47f64..fe7f280 100644
--- a/Namcap/util.py
+++ b/Namcap/util.py
@@ -46,16 +46,23 @@ def _read_carefully(path, readcall):
 		os.chmod(path, statinfo.st_mode)
 	return val
 
-def is_elf(path):
-	"""
-	Given a file path, ensure it exists and peek at the first few bytes
-	to determine if it is an ELF file.
-	"""
-	magic = _read_carefully(path, lambda fd: fd.read(4))
-	if not magic:
-		return False
-	# magic elf header, present in binaries and libraries
-	return magic == b"\x7FELF"
+def _file_has_magic(fileobj, magic_bytes):
+	length = len(magic_bytes)
+	magic = fileobj.read(length)
+	fileobj.seek(0)
+	return magic == magic_bytes
+
+def is_elf(fileobj):
+	"Take file object, peek at the magic bytes to check if ELF file."
+	return _file_has_magic(fileobj, b"\x7fELF")
+
+def is_static(fileobj):
+	"Take file object, peek at the magic bytes to check if static lib."
+	return _file_has_magic(fileobj, b"!<arch>\n")
+
+def is_script(fileobj):
+	"Take file object, peek at the magic bytes to check if script."
+	return _file_has_magic(fileobj, b"#!")
 
 def script_type(path):
 	firstline = _read_carefully(path, lambda fd: fd.readline())