X-Git-Url: https://asedeno.scripts.mit.edu/gitweb/?a=blobdiff_plain;f=diffcore-delta.c;h=e670f8512558c38d9a9d6e754cfc609b042b1195;hb=847d10f56d7853cd0e139a7c2e6ad0ad2de5c464;hp=70bacff83752a6f91b6aa3089374660840e8059a;hpb=6013f17d8852c4b6365e143422e7a85f88efc492;p=git.git

diff --git a/diffcore-delta.c b/diffcore-delta.c
index 70bacff83..e670f8512 100644
--- a/diffcore-delta.c
+++ b/diffcore-delta.c
@@ -5,79 +5,220 @@
 /*
  * Idea here is very simple.
  *
- * We have total of (sz-N+1) N-byte overlapping sequences in buf whose
- * size is sz.  If the same N-byte sequence appears in both source and
- * destination, we say the byte that starts that sequence is shared
- * between them (i.e. copied from source to destination).
+ * Almost all data we are interested in are text, but sometimes we have
+ * to deal with binary data.  So we cut them into chunks delimited by
+ * LF byte, or 64-byte sequence, whichever comes first, and hash them.
  *
- * For each possible N-byte sequence, if the source buffer has more
- * instances of it than the destination buffer, that means the
- * difference are the number of bytes not copied from source to
- * destination.  If the counts are the same, everything was copied
- * from source to destination.  If the destination has more,
- * everything was copied, and destination added more.
+ * For those chunks, if the source buffer has more instances of it
+ * than the destination buffer, that means the difference are the
+ * number of bytes not copied from source to destination.  If the
+ * counts are the same, everything was copied from source to
+ * destination.  If the destination has more, everything was copied,
+ * and destination added more.
  *
  * We are doing an approximation so we do not really have to waste
  * memory by actually storing the sequence.  We just hash them into
  * somewhere around 2^16 hashbuckets and count the occurrences.
- *
- * The length of the sequence is arbitrarily set to 8 for now.
  */
 
-#define HASHBASE 65537 /* next_prime(2^16) */
+/* Wild guess at the initial hash size */
+#define INITIAL_HASH_SIZE 9
+
+/* We leave more room in smaller hash but do not let it
+ * grow to have unused hole too much.
+ */
+#define INITIAL_FREE(sz_log2) ((1<<(sz_log2))*(sz_log2-3)/(sz_log2))
+
+/* A prime rather carefully chosen between 2^16..2^17, so that
+ * HASHBASE < INITIAL_FREE(17).  We want to keep the maximum hashtable
+ * size under the current 2<<17 maximum, which can hold this many
+ * different values before overflowing to hashtable of size 2<<18.
+ */
+#define HASHBASE 107927
+
+struct spanhash {
+	unsigned int hashval;
+	unsigned int cnt;
+};
+struct spanhash_top {
+	int alloc_log2;
+	int free;
+	struct spanhash data[FLEX_ARRAY];
+};
+
+static struct spanhash_top *spanhash_rehash(struct spanhash_top *orig)
+{
+	struct spanhash_top *new;
+	int i;
+	int osz = 1 << orig->alloc_log2;
+	int sz = osz << 1;
+
+	new = xmalloc(sizeof(*orig) + sizeof(struct spanhash) * sz);
+	new->alloc_log2 = orig->alloc_log2 + 1;
+	new->free = INITIAL_FREE(new->alloc_log2);
+	memset(new->data, 0, sizeof(struct spanhash) * sz);
+	for (i = 0; i < osz; i++) {
+		struct spanhash *o = &(orig->data[i]);
+		int bucket;
+		if (!o->cnt)
+			continue;
+		bucket = o->hashval & (sz - 1);
+		while (1) {
+			struct spanhash *h = &(new->data[bucket++]);
+			if (!h->cnt) {
+				h->hashval = o->hashval;
+				h->cnt = o->cnt;
+				new->free--;
+				break;
+			}
+			if (sz <= bucket)
+				bucket = 0;
+		}
+	}
+	free(orig);
+	return new;
+}
 
-static void hash_chars(unsigned char *buf, unsigned long sz, int *count)
+static struct spanhash_top *add_spanhash(struct spanhash_top *top,
+					 unsigned int hashval, int cnt)
 {
-	unsigned int accum1, accum2, i;
-
-	/* an 8-byte shift register made of accum1 and accum2.  New
-	 * bytes come at LSB of accum2, and shifted up to accum1
-	 */
-	for (i = accum1 = accum2 = 0; i < 7; i++, sz--) {
-		accum1 = (accum1 << 8) | (accum2 >> 24);
-		accum2 = (accum2 << 8) | *buf++;
+	int bucket, lim;
+	struct spanhash *h;
+
+	lim = (1 << top->alloc_log2);
+	bucket = hashval & (lim - 1);
+	while (1) {
+		h = &(top->data[bucket++]);
+		if (!h->cnt) {
+			h->hashval = hashval;
+			h->cnt = cnt;
+			top->free--;
+			if (top->free < 0)
+				return spanhash_rehash(top);
+			return top;
+		}
+		if (h->hashval == hashval) {
+			h->cnt += cnt;
+			return top;
+		}
+		if (lim <= bucket)
+			bucket = 0;
 	}
+}
+
+static int spanhash_cmp(const void *a_, const void *b_)
+{
+	const struct spanhash *a = a_;
+	const struct spanhash *b = b_;
+
+	/* A count of zero compares at the end.. */
+	if (!a->cnt)
+		return !b->cnt ? 0 : 1;
+	if (!b->cnt)
+		return -1;
+	return a->hashval < b->hashval ? -1 :
+		a->hashval > b->hashval ? 1 : 0;
+}
+
+static struct spanhash_top *hash_chars(struct diff_filespec *one)
+{
+	int i, n;
+	unsigned int accum1, accum2, hashval;
+	struct spanhash_top *hash;
+	unsigned char *buf = one->data;
+	unsigned int sz = one->size;
+	int is_text = !diff_filespec_is_binary(one);
+
+	i = INITIAL_HASH_SIZE;
+	hash = xmalloc(sizeof(*hash) + sizeof(struct spanhash) * (1<<i));
+	hash->alloc_log2 = i;
+	hash->free = INITIAL_FREE(i);
+	memset(hash->data, 0, sizeof(struct spanhash) * (1<<i));
+
+	n = 0;
+	accum1 = accum2 = 0;
 	while (sz) {
-		accum1 = (accum1 << 8) | (accum2 >> 24);
-		accum2 = (accum2 << 8) | *buf++;
-		/* We want something that hashes permuted byte
-		 * sequences nicely; simpler hash like (accum1 ^
-		 * accum2) does not perform as well.
-		 */
-		i = (accum1 + accum2 * 0x61) % HASHBASE;
-		count[i]++;
+		unsigned int c = *buf++;
+		unsigned int old_1 = accum1;
 		sz--;
+
+		/* Ignore CR in CRLF sequence if text */
+		if (is_text && c == '\r' && sz && *buf == '\n')
+			continue;
+
+		accum1 = (accum1 << 7) ^ (accum2 >> 25);
+		accum2 = (accum2 << 7) ^ (old_1 >> 25);
+		accum1 += c;
+		if (++n < 64 && c != '\n')
+			continue;
+		hashval = (accum1 + accum2 * 0x61) % HASHBASE;
+		hash = add_spanhash(hash, hashval, n);
+		n = 0;
+		accum1 = accum2 = 0;
 	}
+	qsort(hash->data,
+		1ul << hash->alloc_log2,
+		sizeof(hash->data[0]),
+		spanhash_cmp);
+	return hash;
 }
 
-int diffcore_count_changes(void *src, unsigned long src_size,
-			   void *dst, unsigned long dst_size,
+int diffcore_count_changes(struct diff_filespec *src,
+			   struct diff_filespec *dst,
+			   void **src_count_p,
+			   void **dst_count_p,
 			   unsigned long delta_limit,
 			   unsigned long *src_copied,
 			   unsigned long *literal_added)
 {
-	int *src_count, *dst_count, i;
+	struct spanhash *s, *d;
+	struct spanhash_top *src_count, *dst_count;
 	unsigned long sc, la;
 
-	if (src_size < 8 || dst_size < 8)
-		return -1;
-
-	src_count = xcalloc(HASHBASE * 2, sizeof(int));
-	dst_count = src_count + HASHBASE;
-	hash_chars(src, src_size, src_count);
-	hash_chars(dst, dst_size, dst_count);
-
+	src_count = dst_count = NULL;
+	if (src_count_p)
+		src_count = *src_count_p;
+	if (!src_count) {
+		src_count = hash_chars(src);
+		if (src_count_p)
+			*src_count_p = src_count;
+	}
+	if (dst_count_p)
+		dst_count = *dst_count_p;
+	if (!dst_count) {
+		dst_count = hash_chars(dst);
+		if (dst_count_p)
+			*dst_count_p = dst_count;
+	}
 	sc = la = 0;
-	for (i = 0; i < HASHBASE; i++) {
-		if (src_count[i] < dst_count[i]) {
-			la += dst_count[i] - src_count[i];
-			sc += src_count[i];
+
+	s = src_count->data;
+	d = dst_count->data;
+	for (;;) {
+		unsigned dst_cnt, src_cnt;
+		if (!s->cnt)
+			break; /* we checked all in src */
+		while (d->cnt) {
+			if (d->hashval >= s->hashval)
+				break;
+			d++;
 		}
-		else /* i.e. if (dst_count[i] <= src_count[i]) */
-			sc += dst_count[i];
+		src_cnt = s->cnt;
+		dst_cnt = d->hashval == s->hashval ? d->cnt : 0;
+		if (src_cnt < dst_cnt) {
+			la += dst_cnt - src_cnt;
+			sc += src_cnt;
+		}
+		else
+			sc += dst_cnt;
+		s++;
 	}
+
+	if (!src_count_p)
+		free(src_count);
+	if (!dst_count_p)
+		free(dst_count);
 	*src_copied = sc;
 	*literal_added = la;
-	free(src_count);
 	return 0;
 }