From 7867618f873f15a95c22ea899cef07679a8e91a4 Mon Sep 17 00:00:00 2001
From: Emmanuel Marty <emmanuel@fgl.com>
Date: Tue, 25 Jun 2019 11:09:19 +0200
Subject: [PATCH] Fix matchfinder limitation

---
 src/lzsa.c           |  2 +-
 src/matchfinder.c    | 21 ++++++++-------------
 src/shrink_context.h |  6 ++++--
 3 files changed, 13 insertions(+), 16 deletions(-)

diff --git a/src/lzsa.c b/src/lzsa.c
index 1acb2c3..4ab941a 100755
--- a/src/lzsa.c
+++ b/src/lzsa.c
@@ -46,7 +46,7 @@
 #define OPT_RAW         2
 #define OPT_FAVOR_RATIO 4
 
-#define TOOL_VERSION "1.0.1"
+#define TOOL_VERSION "1.0.2"
 
  /*---------------------------------------------------------------------------*/
 
diff --git a/src/matchfinder.c b/src/matchfinder.c
index 86881a9..321927a 100644
--- a/src/matchfinder.c
+++ b/src/matchfinder.c
@@ -192,33 +192,27 @@ int lzsa_find_matches_at(lzsa_compressor *pCompressor, const int nOffset, lzsa_m
    /* Ascend until we reach a visited interval, the root, or a child of the
     * root.  Link unvisited intervals to the current suffix as we go.  */
    while ((super_ref = intervals[ref & POS_MASK]) & LCP_MASK) {
-      intervals[ref & POS_MASK] = nOffset;
+      intervals[ref & POS_MASK] = nOffset | VISITED_FLAG;
       ref = super_ref;
    }
 
    if (super_ref == 0) {
       /* In this case, the current interval may be any of:
        * (1) the root;
-       * (2) an unvisited child of the root;
-       * (3) an interval last visited by suffix 0
-       *
-       * We could avoid the ambiguity with (3) by using an lcp
-       * placeholder value other than 0 to represent "visited", but
-       * it's fastest to use 0.  So we just don't allow matches with
-       * position 0.  */
+       * (2) an unvisited child of the root */
 
       if (ref != 0)  /* Not the root?  */
-         intervals[ref & POS_MASK] = nOffset;
+         intervals[ref & POS_MASK] = nOffset | VISITED_FLAG;
       return 0;
    }
 
    /* Ascend indirectly via pos_data[] links.  */
-   match_pos = super_ref;
+   match_pos = super_ref & EXCL_VISITED_MASK;
    matchptr = pMatches;
    for (;;) {
       while ((super_ref = pos_data[match_pos]) > ref)
-         match_pos = intervals[super_ref & POS_MASK];
-      intervals[ref & POS_MASK] = nOffset;
+         match_pos = intervals[super_ref & POS_MASK] & EXCL_VISITED_MASK;
+      intervals[ref & POS_MASK] = nOffset | VISITED_FLAG;
       pos_data[match_pos] = ref;
 
       if ((matchptr - pMatches) < nMaxMatches) {
@@ -234,7 +228,7 @@ int lzsa_find_matches_at(lzsa_compressor *pCompressor, const int nOffset, lzsa_m
       if (super_ref == 0)
          break;
       ref = super_ref;
-      match_pos = intervals[ref & POS_MASK];
+      match_pos = intervals[ref & POS_MASK] & EXCL_VISITED_MASK;
    }
 
    return (int)(matchptr - pMatches);
@@ -263,6 +257,7 @@ void lzsa_skip_matches(lzsa_compressor *pCompressor, const int nStartOffset, con
  * the optimizer to look at.
  *
  * @param pCompressor compression context
+ * @param pInWindow pointer to input data window (previously compressed bytes + bytes to compress)
  * @param nStartOffset current offset in input window (typically the number of previously compressed bytes)
  * @param nEndOffset offset to end finding matches at (typically the size of the total input window in bytes
  */
diff --git a/src/shrink_context.h b/src/shrink_context.h
index fa24e60..7360044 100644
--- a/src/shrink_context.h
+++ b/src/shrink_context.h
@@ -39,11 +39,13 @@
 extern "C" {
 #endif
 
-#define LCP_BITS 15
+#define LCP_BITS 14
 #define LCP_MAX (1U<<(LCP_BITS - 1))
-#define LCP_SHIFT (32-LCP_BITS)
+#define LCP_SHIFT (31-LCP_BITS)
 #define LCP_MASK (((1U<<LCP_BITS) - 1) << LCP_SHIFT)
 #define POS_MASK ((1U<<LCP_SHIFT) - 1)
+#define VISITED_FLAG 0x80000000
+#define EXCL_VISITED_MASK  0x7fffffff
 
 #define NMATCHES_PER_OFFSET 8
 #define MATCHES_PER_OFFSET_SHIFT 3