dblink._collision_protect: Detect internal collisions.

Implement detection of internal collisions (between files of the same package,
located in separate directories in the installation image (${D}) corresponding
to merged directories in the target filesystem (${ROOT})).

This provides protection against overwriting some files when performing merging
of files from ${D} to ${ROOT} in some filesystem layouts (such as /-merged layout
or /usr-merged layout).

Internal collisions between identical files are silently ignored.

Bug: https://bugs.gentoo.org/690484
Signed-off-by: Arfrever Frehtes Taifersar Arahesis <Arfrever@Apache.Org>
Signed-off-by: Zac Medico <zmedico@gentoo.org>

1 files changed, 103 insertions, 0 deletions

diff --git a/lib/portage/util/_compare_files.py b/lib/portage/util/_compare_files.py
new file mode 100644
index 000000000..bd993e501
--- /dev/null
+++ b/lib/portage/util/_compare_files.py
@@ -0,0 +1,103 @@
+# Copyright 2019 Gentoo Authors
+# Distributed under the terms of the GNU General Public License v2
+
+__all__ = ["compare_files"]
+
+import io
+import os
+import stat
+import sys
+
+from portage import _encodings
+from portage import _unicode_encode
+from portage.util._xattr import xattr
+
+def compare_files(file1, file2, skipped_types=()):
+	"""
+	Compare metadata and contents of two files.
+
+	@param file1: File 1
+	@type file1: str
+	@param file2: File 2
+	@type file2: str
+	@param skipped_types: Tuple of strings specifying types of properties excluded from comparison.
+		Supported strings: type, mode, owner, group, device_number, xattr, atime, mtime, ctime, size, content
+	@type skipped_types: tuple of str
+	@rtype: tuple of str
+	@return: Tuple of strings specifying types of properties different between compared files
+	"""
+
+	file1_stat = os.lstat(_unicode_encode(file1, encoding=_encodings["fs"], errors="strict"))
+	file2_stat = os.lstat(_unicode_encode(file2, encoding=_encodings["fs"], errors="strict"))
+
+	differences = []
+
+	if (file1_stat.st_dev, file1_stat.st_ino) == (file2_stat.st_dev, file2_stat.st_ino):
+		return ()
+
+	if "type" not in skipped_types and stat.S_IFMT(file1_stat.st_mode) != stat.S_IFMT(file2_stat.st_mode):
+		differences.append("type")
+	if "mode" not in skipped_types and stat.S_IMODE(file1_stat.st_mode) != stat.S_IMODE(file2_stat.st_mode):
+		differences.append("mode")
+	if "owner" not in skipped_types and file1_stat.st_uid != file2_stat.st_uid:
+		differences.append("owner")
+	if "group" not in skipped_types and file1_stat.st_gid != file2_stat.st_gid:
+		differences.append("group")
+	if "device_number" not in skipped_types and file1_stat.st_rdev != file2_stat.st_rdev:
+		differences.append("device_number")
+
+	if "xattr" not in skipped_types and sorted(xattr.get_all(file1, nofollow=True)) != sorted(xattr.get_all(file2, nofollow=True)):
+		differences.append("xattr")
+
+	if sys.hexversion >= 0x3030000:
+		if "atime" not in skipped_types and file1_stat.st_atime_ns != file2_stat.st_atime_ns:
+			differences.append("atime")
+		if "mtime" not in skipped_types and file1_stat.st_mtime_ns != file2_stat.st_mtime_ns:
+			differences.append("mtime")
+		if "ctime" not in skipped_types and file1_stat.st_ctime_ns != file2_stat.st_ctime_ns:
+			differences.append("ctime")
+	else:
+		if "atime" not in skipped_types and file1_stat.st_atime != file2_stat.st_atime:
+			differences.append("atime")
+		if "mtime" not in skipped_types and file1_stat.st_mtime != file2_stat.st_mtime:
+			differences.append("mtime")
+		if "ctime" not in skipped_types and file1_stat.st_ctime != file2_stat.st_ctime:
+			differences.append("ctime")
+
+	if "type" in differences:
+		pass
+	elif file1_stat.st_size != file2_stat.st_size:
+		if "size" not in skipped_types:
+			differences.append("size")
+		if "content" not in skipped_types:
+			differences.append("content")
+	else:
+		if "content" not in skipped_types:
+			if stat.S_ISLNK(file1_stat.st_mode):
+				file1_stream = io.BytesIO(os.readlink(_unicode_encode(file1,
+									encoding=_encodings["fs"],
+									errors="strict")))
+			else:
+				file1_stream = open(_unicode_encode(file1,
+							encoding=_encodings["fs"],
+							errors="strict"), "rb")
+			if stat.S_ISLNK(file2_stat.st_mode):
+				file2_stream = io.BytesIO(os.readlink(_unicode_encode(file2,
+									encoding=_encodings["fs"],
+									errors="strict")))
+			else:
+				file2_stream = open(_unicode_encode(file2,
+							encoding=_encodings["fs"],
+							errors="strict"), "rb")
+			while True:
+				file1_content = file1_stream.read(4096)
+				file2_content = file2_stream.read(4096)
+				if file1_content != file2_content:
+					differences.append("content")
+					break
+				if not file1_content or not file2_content:
+					break
+			file1_stream.close()
+			file2_stream.close()
+
+	return tuple(differences)