X-Git-Url: https://mattmccutchen.net/rsync/rsync.git/blobdiff_plain/51b2ff03b3f45b360b75109148cd0606621089bc..3f26945cb1a602d3f838507ac77079f649bf9086:/match.c

diff --git a/match.c b/match.c
index 229c9dd6..f8a1a8f9 100644
--- a/match.c
+++ b/match.c
@@ -91,8 +91,7 @@ static void build_hash_table(struct sum_struct *s)
 static OFF_T last_match;
 
 
-/**
- * Transmit a literal and/or match token.
+/* Transmit a literal and/or match token.
  *
  * This delightfully-named function is called either when we find a
  * match and need to transmit all the unmatched data leading up to it,
@@ -100,9 +99,9 @@ static OFF_T last_match;
  * transmit it.  As a result of this second case, it is called even if
  * we have not matched at all!
  *
- * @param i If >0, the number of a matched token.  If 0, indicates we
- * have only literal data.
- **/
+ * If i >= 0, the number of a matched token.  If < 0, indicates we have
+ * only literal data.  A -1 will send a 0-token-int too, and a -2 sends
+ * only literal data, w/o any token-int. */
 static void matched(int f, struct sum_struct *s, struct map_struct *buf,
 		    OFF_T offset, int32 i)
 {
@@ -235,24 +234,47 @@ static void hash_search(int f,struct sum_struct *s,
 			 * one with an identical offset, so we prefer that over
 			 * the adjacent want_i optimization. */
 			if (updating_basis_file) {
-				/* All the sender's chunks start at blength boundaries. */
+				/* All the generator's chunks start at blength boundaries. */
 				while (aligned_offset < offset)
 					aligned_offset += s->blength;
-				if (offset == aligned_offset) {
+				if (offset == aligned_offset
+				 || (sum == 0 && l == s->blength && aligned_offset + l <= len)) {
 					int32 i2;
 					for (i2 = i; i2 >= 0; i2 = s->sums[i2].chain) {
-						if (s->sums[i2].offset != offset)
+						if (s->sums[i2].offset != aligned_offset)
 							continue;
 						if (i2 != i) {
 							if (sum != s->sums[i2].sum1
+							 || l != s->sums[i2].len
 							 || memcmp(sum2, s->sums[i2].sum2, s->s2length) != 0)
 								break;
 							i = i2;
 						}
-						/* This chunk was at the same offset on
-						 * both the sender and the receiver. */
-						s->sums[i].flags |= SUMFLG_SAME_OFFSET;
 						want_i = i;
+						if (offset != aligned_offset) {
+							/* We've matched some zeros in a spot that is also zeros
+							 * further along in the basis file, if we find zeros ahead
+							 * in the sender's file, we'll output enough literal data
+							 * to re-align with the basis file, and get back to seeking
+							 * instead of writing. */
+							backup = (int32)(aligned_offset - last_match);
+							if (backup < 0)
+								backup = 0;
+							map = (schar *)map_ptr(buf, aligned_offset - backup, l + backup)
+							    + backup;
+							sum = get_checksum1((char *)map, l);
+							if (sum != s->sums[i2].sum1)
+								break;
+							get_checksum2((char *)map, l, sum2);
+							if (memcmp(sum2, s->sums[i2].sum2, s->s2length) != 0)
+								break;
+							/* OK, we have a re-alignment match.  Bump the offset
+							 * forward to the new match point. */
+							offset = aligned_offset;
+						}
+						/* This chunk remained in the same spot in the old and new file. */
+						s->sums[i].flags |= SUMFLG_SAME_OFFSET;
+						break;
 					}
 				}
 			}