From 935717373b2fcf61eae02d4ba250bbd2bb9116bc Mon Sep 17 00:00:00 2001
From: Moinak Ghosh <moinakg@gmail.com>
Date: Sun, 30 Mar 2014 17:35:21 +0530
Subject: [PATCH] Capability to list offset and length of each block when
 deduplication for external use.

---
 README.md           | 3 +++
 pcompress.c         | 6 +++++-
 pcompress.h         | 1 +
 rabin/rabin_dedup.c | 7 +++++++
 rabin/rabin_dedup.h | 1 +
 5 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/README.md b/README.md
index c08699c..7e0f028 100644
--- a/README.md
+++ b/README.md
@@ -361,6 +361,9 @@ Advanced usage
 
        -M       Display memory allocator statistics.
        -C       Display compression statistics.
+       -CC      Display compression statistics and print the offset and length of each
+                variable length dedupe block if variable block deduplication is being
+                used. This has no effect for fixed block deduplication.
 
 Environment Variables
 =====================
diff --git a/pcompress.c b/pcompress.c
index 379cc6c..260bd5a 100644
--- a/pcompress.c
+++ b/pcompress.c
@@ -2243,6 +2243,7 @@ start_compress(pc_ctx_t *pctx, const char *filename, uint64_t chunksize, int lev
 				COMP_BAIL;
 			}
 
+			tdat->rctx->show_chunks = pctx->show_chunks;
 			tdat->rctx->index_sem = &(tdat->index_sem);
 			tdat->rctx->id = i;
 		}
@@ -2906,7 +2907,10 @@ init_pc_context(pc_ctx_t *pctx, int argc, char *argv[])
 			break;
 
 		    case 'C':
-			pctx->hide_cmp_stats = 0;
+			if (pctx->hide_cmp_stats)
+				pctx->hide_cmp_stats = 0;
+			else
+				pctx->show_chunks = 1;
 			break;
 
 		    case 'D':
diff --git a/pcompress.h b/pcompress.h
index 17b9c45..27e5688 100644
--- a/pcompress.h
+++ b/pcompress.h
@@ -198,6 +198,7 @@ typedef struct pc_ctx {
 	int nthreads;
 	int hide_mem_stats;
 	int hide_cmp_stats;
+	int show_chunks;
 	int enable_rabin_scan;
 	int enable_rabin_global;
 	int enable_delta_encode;
diff --git a/rabin/rabin_dedup.c b/rabin/rabin_dedup.c
index 923eed1..fe8377b 100755
--- a/rabin/rabin_dedup.c
+++ b/rabin/rabin_dedup.c
@@ -293,6 +293,7 @@ create_dedupe_context(uint64_t chunksize, uint64_t real_chunksize, int rab_blk_s
 	ctx->deltac_min_distance = props->deltac_min_distance;
 	ctx->pagesize = sysconf(_SC_PAGE_SIZE);
 	ctx->similarity_cksums = NULL;
+	ctx->show_chunks = 0;
 	if (arc) {
 		arc->pagesize = ctx->pagesize;
 		if (rab_blk_sz < 3)
@@ -679,6 +680,9 @@ dedupe_compress(dedupe_context_t *ctx, uchar_t *buf, uint64_t *size, uint64_t of
 				ctx->g_blocks[blknum].offset = last_offset;
 			}
 			DEBUG_STAT_EN(if (length >= ctx->rabin_poly_max_block_size) ++max_count);
+			if (ctx->show_chunks) {
+				fprintf(stderr, "Block offset: %" PRIu64 ", length: %u\n", last_offset, length);
+			}
 
 			/*
 			 * Reset the heap structure and find the K min values if Delta Compression
@@ -726,6 +730,9 @@ dedupe_compress(dedupe_context_t *ctx, uchar_t *buf, uint64_t *size, uint64_t of
 			ctx->g_blocks[blknum].length = length;
 			ctx->g_blocks[blknum].offset = last_offset;
 		}
+		if (ctx->show_chunks) {
+			fprintf(stderr, "Block offset: %" PRIu64 ", length: %u\n", last_offset, length);
+		}
 
 		if (ctx->delta_flag) {
 			uint64_t cur_sketch;
diff --git a/rabin/rabin_dedup.h b/rabin/rabin_dedup.h
index 9fbb0a4..ce4f05f 100644
--- a/rabin/rabin_dedup.h
+++ b/rabin/rabin_dedup.h
@@ -188,6 +188,7 @@ typedef struct {
 	uint32_t pagesize;
 	int out_fd;
 	int id;
+	int show_chunks; // Debug display of chunks (offset, length)
 } dedupe_context_t;
 
 extern dedupe_context_t *create_dedupe_context(uint64_t chunksize, uint64_t real_chunksize,