From 683c3e48b5704f1e3c293825950588f4a63c62fc Mon Sep 17 00:00:00 2001
From: Moinak Ghosh <moinakg@gmail.com>
Date: Wed, 1 Jan 2014 19:44:58 +0530
Subject: [PATCH] Detect some DICOM formats and use BSC for DICOM data.

---
 adaptive_compress.c  | 12 +++++++++---
 archive/pc_archive.c | 16 ++++++++++++++++
 utils/utils.h        |  3 ++-
 3 files changed, 27 insertions(+), 4 deletions(-)

diff --git a/adaptive_compress.c b/adaptive_compress.c
index edbb6d2..b2f4b71 100644
--- a/adaptive_compress.c
+++ b/adaptive_compress.c
@@ -215,12 +215,15 @@ adapt_deinit(void **data)
 	return (rv);
 }
 
+/*
+ * Identify the types that BSC can compress better than others.
+ */
 int
 is_bsc_type(int btype)
 {
 	int stype = PC_SUBTYPE(btype);
 	return ((stype == TYPE_MARKUP) | (stype == TYPE_BMP) | (stype == TYPE_DNA_SEQ) |
-	    (stype == TYPE_MP4) | (stype == TYPE_FLAC) | (stype == TYPE_AVI));
+	    (stype == TYPE_MP4) | (stype == TYPE_FLAC) | (stype == TYPE_AVI) | (stype == TYPE_DICOM));
 }
 
 int
@@ -237,8 +240,8 @@ adapt_compress(void *src, uint64_t srclen, void *dst,
 		double tagcnt, pct_tag;
 		uchar_t cur_byte, prev_byte;
 		/*
-		* Count number of 8-bit binary bytes and XML tags in source.
-		*/
+		 * Count number of 8-bit binary bytes and XML tags in source.
+		 */
 		tot8b = 0;
 		tag1 = 0;
 		tag2 = 0;
@@ -255,6 +258,9 @@ adapt_compress(void *src, uint64_t srclen, void *dst,
 				prev_byte = cur_byte;
 		}
 
+		/*
+		 * Heuristics for detecting BINARY vs generic TEXT vs XML data.
+		 */
 		tot8b /= 0x80;
 		tagcnt = tag1 + tag2 + tag3;
 		pct_tag = tagcnt / (double)srclen;
diff --git a/archive/pc_archive.c b/archive/pc_archive.c
index 01264b1..211a2fc 100644
--- a/archive/pc_archive.c
+++ b/archive/pc_archive.c
@@ -1493,6 +1493,22 @@ detect_type_by_data(uchar_t *buf, size_t len)
 		return (TYPE_BINARY|TYPE_ARCHIVE_TAR);
 	if (memcmp(buf, "%PDF-", 5) == 0)
 		return (TYPE_BINARY|TYPE_PDF);
+
+	// Try to detect DICOM medical image file. BSC compresses these better.
+	if (len > 127) {
+		size_t i;
+
+		// DICOM files should have either DICM or ISO_IR within the first 128 bytes
+		for (i = 0; i < 128; i++) {
+			if (buf[i] == 'D')
+				if (memcmp(&buf[i], "DICM", 4) == 0)
+					return (TYPE_BINARY|TYPE_DICOM);
+			if (buf[i] == 'I')
+				if (memcmp(&buf[i], "ISO_IR ", 7) == 0)
+					return (TYPE_BINARY|TYPE_DICOM);
+		}
+	}
+
 	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/utils/utils.h b/utils/utils.h
index 358d07d..17c31d5 100644
--- a/utils/utils.h
+++ b/utils/utils.h
@@ -274,7 +274,8 @@ typedef enum {
 	TYPE_BMP = 200,
 	TYPE_TIFF = 208,
 	TYPE_PDF = 216,
-	TYPE_ARCHIVE_TAR = 224
+	TYPE_ARCHIVE_TAR = 224,
+	TYPE_DICOM = 232
 } data_type_t;
 
 /*