From 5521955a944fb6017d9ec4643f4dc48d21d1527c Mon Sep 17 00:00:00 2001
From: Moinak Ghosh <moinakg@gmail.com>
Date: Wed, 18 Dec 2013 23:00:39 +0530
Subject: [PATCH] Detect AR archives and set the type. Re-use a less common
 type code for AR. Use Dispack generically for all executables and AR
 archives.

---
 archive/pc_archive.c       | 2 ++
 pcompress.c                | 5 ++++-
 utils/phash/extensions.h   | 4 ++--
 utils/phash/extensions.txt | 4 ++--
 utils/utils.h              | 2 +-
 5 files changed, 11 insertions(+), 6 deletions(-)

diff --git a/archive/pc_archive.c b/archive/pc_archive.c
index 8708950..d118d1d 100644
--- a/archive/pc_archive.c
+++ b/archive/pc_archive.c
@@ -1467,6 +1467,8 @@ detect_type_by_data(uchar_t *buf, size_t len)
 	// At least a few bytes.
 	if (len < 16) return (TYPE_UNKNOWN);
 
+	if (memcmp(buf, "!<arch>\n", 8) == 0)
+		return (TYPE_BINARY|TYPE_ARCHIVE_AR);
 	if (U32_P(buf) == ELFINT) {  // Regular ELF, check for 32/64-bit, core dump
 		if (*(buf + 16) != 4) {
 			if (*(buf + 4) == 2) {
diff --git a/pcompress.c b/pcompress.c
index 4cd9625..6eed6e3 100644
--- a/pcompress.c
+++ b/pcompress.c
@@ -226,8 +226,11 @@ preproc_compress(pc_ctx_t *pctx, compress_func_ptr cmp_func, void *src, uint64_t
 	/*
 	 * If Dispack is enabled it has to be done first since Dispack analyses the
 	 * x86 instruction stream in the raw data.
+	 * AR archives are typically static libraries. So we Dispack them unconditionally.
+	 * TODO: Is this too much to assume in the generic case? Can we look inside ar archives?
 	 */
-	if (pctx->dispack_preprocess && stype == TYPE_EXE32) {
+	if (pctx->dispack_preprocess && (stype == TYPE_EXE32 || stype == TYPE_EXE64 ||
+	    stype == TYPE_ARCHIVE_AR)) {
 		_dstlen = fromlen;
 		result = dispack_encode((uchar_t *)from, fromlen, to, &_dstlen);
 		if (result != -1) {
diff --git a/utils/phash/extensions.h b/utils/phash/extensions.h
index e4c239b..7275e5e 100644
--- a/utils/phash/extensions.h
+++ b/utils/phash/extensions.h
@@ -121,8 +121,8 @@ struct ext_entry {
 	{"arc"	, TYPE_BINARY|TYPE_COMPRESSED|TYPE_COMPRESSED_ARC, 3},
 	{"jar"	, TYPE_BINARY|TYPE_COMPRESSED|TYPE_COMPRESSED_GZ, 3},
 	{"lz"	, TYPE_BINARY|TYPE_COMPRESSED|TYPE_COMPRESSED_LZ, 2},
-	{"lzh"	, TYPE_BINARY|TYPE_COMPRESSED|TYPE_COMPRESSED_LH, 3},
-	{"lha"	, TYPE_BINARY|TYPE_COMPRESSED|TYPE_COMPRESSED_LH, 3},
+	{"lzh"	, TYPE_BINARY|TYPE_COMPRESSED, 3},
+	{"lha"	, TYPE_BINARY|TYPE_COMPRESSED, 3},
 	{"lzma"	, TYPE_BINARY|TYPE_COMPRESSED|TYPE_COMPRESSED_LZMA, 4},
 	{"lzo"	, TYPE_BINARY|TYPE_COMPRESSED|TYPE_COMPRESSED_LZO, 3},
 	{"dmg"	, TYPE_BINARY, 3},
diff --git a/utils/phash/extensions.txt b/utils/phash/extensions.txt
index 22a8943..323af6c 100644
--- a/utils/phash/extensions.txt
+++ b/utils/phash/extensions.txt
@@ -111,8 +111,8 @@ arj,TYPE_BINARY|TYPE_COMPRESSED|TYPE_COMPRESSED_ARJ
 arc,TYPE_BINARY|TYPE_COMPRESSED|TYPE_COMPRESSED_ARC
 jar,TYPE_BINARY|TYPE_COMPRESSED|TYPE_COMPRESSED_GZ
 lz,TYPE_BINARY|TYPE_COMPRESSED|TYPE_COMPRESSED_LZ
-lzh,TYPE_BINARY|TYPE_COMPRESSED|TYPE_COMPRESSED_LH
-lha,TYPE_BINARY|TYPE_COMPRESSED|TYPE_COMPRESSED_LH
+lzh,TYPE_BINARY|TYPE_COMPRESSED
+lha,TYPE_BINARY|TYPE_COMPRESSED
 lzma,TYPE_BINARY|TYPE_COMPRESSED|TYPE_COMPRESSED_LZMA
 lzo,TYPE_BINARY|TYPE_COMPRESSED|TYPE_COMPRESSED_LZO
 dmg,TYPE_BINARY
diff --git a/utils/utils.h b/utils/utils.h
index dbac54f..bdf5021 100644
--- a/utils/utils.h
+++ b/utils/utils.h
@@ -256,7 +256,7 @@ typedef enum {
 	TYPE_COMPRESSED_ZIP = 56,
 	TYPE_COMPRESSED_ARJ = 64,
 	TYPE_COMPRESSED_ARC = 72,
-	TYPE_COMPRESSED_LH = 80,
+	TYPE_ARCHIVE_AR = 80,
 	TYPE_COMPRESSED_LZMA = 88,
 	TYPE_COMPRESSED_LZO = 96,
 	TYPE_AVI = 104,