vendor/libgit2/src/hashsig.c in rugged-0.22.2 vs vendor/libgit2/src/hashsig.c in rugged-0.23.0b1

@@ -32,12 +32,12 @@
 } hashsig_heap;
 
 struct git_hashsig {
 	hashsig_heap mins;
 	hashsig_heap maxs;
+	size_t lines;
 	git_hashsig_option_t opt;
-	int considered;
 };
 
 #define HEAP_LCHILD_OF(I) (((I)<<1)+1)
 #define HEAP_RCHILD_OF(I) (((I)<<1)+2)
 #define HEAP_PARENT_OF(I) (((I)-1)>>1)
@@ -133,29 +133,27 @@
 static void hashsig_in_progress_init(
 	hashsig_in_progress *prog, git_hashsig *sig)
 {
 	int i;
 
-	switch (sig->opt) {
-	case GIT_HASHSIG_IGNORE_WHITESPACE:
+	/* no more than one can be set */
+	assert(!(sig->opt & GIT_HASHSIG_IGNORE_WHITESPACE) ||
+		   !(sig->opt & GIT_HASHSIG_SMART_WHITESPACE));
+
+	if (sig->opt & GIT_HASHSIG_IGNORE_WHITESPACE) {
 		for (i = 0; i < 256; ++i)
 			prog->ignore_ch[i] = git__isspace_nonlf(i);
 		prog->use_ignores = 1;
-		break;
-	case GIT_HASHSIG_SMART_WHITESPACE:
+	} else if (sig->opt & GIT_HASHSIG_SMART_WHITESPACE) {
 		for (i = 0; i < 256; ++i)
 			prog->ignore_ch[i] = git__isspace(i);
 		prog->use_ignores = 1;
-		break;
-	default:
+	} else {
 		memset(prog, 0, sizeof(*prog));
-		break;
 	}
 }
 
-#define HASHSIG_IN_PROGRESS_INIT { 1 }
-
 static int hashsig_add_hashes(
 	git_hashsig *sig,
 	const uint8_t *data,
 	size_t size,
 	hashsig_in_progress *prog)
@@ -172,36 +170,37 @@
 			ch = *scan;
 
 			if (use_ignores)
 				for (; scan < end && git__isspace_nonlf(ch); ch = *scan)
 					++scan;
-			else if (sig->opt != GIT_HASHSIG_NORMAL)
+			else if (sig->opt &
+					 (GIT_HASHSIG_IGNORE_WHITESPACE | GIT_HASHSIG_SMART_WHITESPACE))
 				for (; scan < end && ch == '\r'; ch = *scan)
 					++scan;
 
 			/* peek at next character to decide what to do next */
-			if (sig->opt == GIT_HASHSIG_SMART_WHITESPACE)
+			if (sig->opt & GIT_HASHSIG_SMART_WHITESPACE)
 				use_ignores = (ch == '\n');
 
 			if (scan >= end)
 				break;
 			++scan;
 
 			/* check run terminator */
-			if (ch == '\n' || ch == '\0')
+			if (ch == '\n' || ch == '\0') {
+				sig->lines++;
 				break;
+			}
 
 			++len;
 			HASHSIG_HASH_MIX(state, ch);
 		}
 
 		if (len > 0) {
 			hashsig_heap_insert(&sig->mins, (hashsig_t)state);
 			hashsig_heap_insert(&sig->maxs, (hashsig_t)state);
 
-			sig->considered++;
-
 			while (scan < end && (*scan == '\n' || !*scan))
 				++scan;
 		}
 	}
 
@@ -210,11 +209,12 @@
 	return 0;
 }
 
 static int hashsig_finalize_hashes(git_hashsig *sig)
 {
-	if (sig->mins.size < HASHSIG_HEAP_MIN_SIZE) {
+	if (sig->mins.size < HASHSIG_HEAP_MIN_SIZE &&
+		!(sig->opt & GIT_HASHSIG_ALLOW_SMALL_FILES)) {
 		giterr_set(GITERR_INVALID,
 			"File too small for similarity signature calculation");
 		return GIT_EBUFS;
 	}
 
@@ -334,9 +334,21 @@
 	return HASHSIG_SCALE * (matches * 2) / (a->size + b->size);
 }
 
 int git_hashsig_compare(const git_hashsig *a, const git_hashsig *b)
 {
+	/* if we have no elements in either file then each file is either
+	 * empty or blank.  if we're ignoring whitespace then the files are
+	 * similar, otherwise they're dissimilar.
+	 */
+	if (a->mins.size == 0 && b->mins.size == 0) {
+		if ((!a->lines && !b->lines) ||
+			(a->opt & GIT_HASHSIG_IGNORE_WHITESPACE))
+			return HASHSIG_SCALE;
+		else
+			return 0;
+	}
+
 	/* if we have fewer than the maximum number of elements, then just use
 	 * one array since the two arrays will be the same
 	 */
 	if (a->mins.size < HASHSIG_HEAP_SIZE)
 		return hashsig_heap_compare(&a->mins, &b->mins);