rugged 0.19.0 → 0.28.0

data/vendor/libgit2/src/hashsig.c

@@ -4,7 +4,10 @@
  * This file is part of libgit2, distributed under the GNU GPL v2 with
  * a Linking Exception. For full terms see the included COPYING file.
  */
-#include "hashsig.h"
+
+#include "common.h"
+
+#include "git2/sys/hashsig.h"
 #include "fileops.h"
 #include "util.h"
 
@@ -13,12 +16,15 @@ typedef uint64_t hashsig_state;
 
 #define HASHSIG_SCALE 100
 
-#define HASHSIG_HASH_WINDOW 32
-#define HASHSIG_HASH_START	0
+#define HASHSIG_MAX_RUN 80
+#define HASHSIG_HASH_START	0x012345678ABCDEF0LL
 #define HASHSIG_HASH_SHIFT  5
-#define HASHSIG_HASH_MASK   0x7FFFFFFF
+
+#define HASHSIG_HASH_MIX(S,CH) \
+	(S) = ((S) << HASHSIG_HASH_SHIFT) - (S) + (hashsig_state)(CH)
 
 #define HASHSIG_HEAP_SIZE ((1 << 7) - 1)
+#define HASHSIG_HEAP_MIN_SIZE 4
 
 typedef int (*hashsig_cmp)(const void *a, const void *b, void *);
 
@@ -28,23 +34,15 @@ typedef struct {
 	hashsig_t values[HASHSIG_HEAP_SIZE];
 } hashsig_heap;
 
-typedef struct {
-	hashsig_state state, shift_n;
-	char window[HASHSIG_HASH_WINDOW];
-	int win_len, win_pos, saw_lf;
-} hashsig_in_progress;
-
-#define HASHSIG_IN_PROGRESS_INIT { HASHSIG_HASH_START, 1, {0}, 0, 0, 1 }
-
 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)*2)+1)
-#define HEAP_RCHILD_OF(I) (((I)*2)+2)
+#define HEAP_LCHILD_OF(I) (((I)<<1)+1)
+#define HEAP_RCHILD_OF(I) (((I)<<1)+2)
 #define HEAP_PARENT_OF(I) (((I)-1)>>1)
 
 static void hashsig_heap_init(hashsig_heap *h, hashsig_cmp cmp)
@@ -115,136 +113,111 @@ static void hashsig_heap_sort(hashsig_heap *h)
 
 static void hashsig_heap_insert(hashsig_heap *h, hashsig_t val)
 {
-	/* if heap is full, pop top if new element should replace it */
-	if (h->size == h->asize && h->cmp(&val, &h->values[0], NULL) > 0) {
-		h->size--;
-		h->values[0] = h->values[h->size];
-		hashsig_heap_down(h, 0);
-	}
-
 	/* if heap is not full, insert new element */
 	if (h->size < h->asize) {
 		h->values[h->size++] = val;
 		hashsig_heap_up(h, h->size - 1);
 	}
-}
 
-GIT_INLINE(bool) hashsig_include_char(
-	char ch, git_hashsig_option_t opt, int *saw_lf)
-{
-	if ((opt & GIT_HASHSIG_IGNORE_WHITESPACE) && git__isspace(ch))
-		return false;
-
-	if (opt & GIT_HASHSIG_SMART_WHITESPACE) {
-		if (ch == '\r' || (*saw_lf && git__isspace(ch)))
-			return false;
-
-		*saw_lf = (ch == '\n');
+	/* if heap is full, pop top if new element should replace it */
+	else if (h->cmp(&val, &h->values[0], NULL) > 0) {
+		h->size--;
+		h->values[0] = h->values[h->size];
+		hashsig_heap_down(h, 0);
 	}
 
-	return true;
 }
 
-static void hashsig_initial_window(
-	git_hashsig *sig,
-	const char **data,
-	size_t size,
-	hashsig_in_progress *prog)
-{
-	hashsig_state state, shift_n;
-	int win_len;
-	const char *scan, *end;
-
-	/* init until we have processed at least HASHSIG_HASH_WINDOW data */
-
-	if (prog->win_len >= HASHSIG_HASH_WINDOW)
-		return;
-
-	state   = prog->state;
-	win_len = prog->win_len;
-	shift_n = prog->shift_n;
-
-	scan = *data;
-	end  = scan + size;
-
-	while (scan < end && win_len < HASHSIG_HASH_WINDOW) {
-		char ch = *scan++;
-
-		if (!hashsig_include_char(ch, sig->opt, &prog->saw_lf))
-			continue;
-
-		state = (state * HASHSIG_HASH_SHIFT + ch) & HASHSIG_HASH_MASK;
-
-		if (!win_len)
-			shift_n = 1;
-		else
-			shift_n = (shift_n * HASHSIG_HASH_SHIFT) & HASHSIG_HASH_MASK;
-
-		prog->window[win_len++] = ch;
-	}
-
-	/* insert initial hash if we just finished */
+typedef struct {
+	int use_ignores;
+	uint8_t ignore_ch[256];
+} hashsig_in_progress;
 
-	if (win_len == HASHSIG_HASH_WINDOW) {
-		hashsig_heap_insert(&sig->mins, (hashsig_t)state);
-		hashsig_heap_insert(&sig->maxs, (hashsig_t)state);
-		sig->considered = 1;
+static void hashsig_in_progress_init(
+	hashsig_in_progress *prog, git_hashsig *sig)
+{
+	int i;
+
+	/* 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;
+	} 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;
+	} else {
+		memset(prog, 0, sizeof(*prog));
 	}
-
-	prog->state   = state;
-	prog->win_len = win_len;
-	prog->shift_n = shift_n;
-
-	*data = scan;
 }
 
 static int hashsig_add_hashes(
 	git_hashsig *sig,
-	const char *data,
+	const uint8_t *data,
 	size_t size,
 	hashsig_in_progress *prog)
 {
-	const char *scan = data, *end = data + size;
-	hashsig_state state, shift_n, rmv;
-
-	if (prog->win_len < HASHSIG_HASH_WINDOW)
-		hashsig_initial_window(sig, &scan, size, prog);
-
-	state   = prog->state;
-	shift_n = prog->shift_n;
-
-	/* advance window, adding new chars and removing old */
-
-	for (; scan < end; ++scan) {
-		char ch = *scan;
-
-		if (!hashsig_include_char(ch, sig->opt, &prog->saw_lf))
-			continue;
-
-		rmv = shift_n * prog->window[prog->win_pos];
-
-		state = (state - rmv) & HASHSIG_HASH_MASK;
-		state = (state * HASHSIG_HASH_SHIFT) & HASHSIG_HASH_MASK;
-		state = (state + ch) & HASHSIG_HASH_MASK;
+	const uint8_t *scan = data, *end = data + size;
+	hashsig_state state = HASHSIG_HASH_START;
+	int use_ignores = prog->use_ignores, len;
+	uint8_t ch;
+
+	while (scan < end) {
+		state = HASHSIG_HASH_START;
+
+		for (len = 0; scan < end && len < HASHSIG_MAX_RUN; ) {
+			ch = *scan;
+
+			if (use_ignores)
+				for (; scan < end && git__isspace_nonlf(ch); ch = *scan)
+					++scan;
+			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)
+				use_ignores = (ch == '\n');
+
+			if (scan >= end)
+				break;
+			++scan;
+
+			/* check run terminator */
+			if (ch == '\n' || ch == '\0') {
+				sig->lines++;
+				break;
+			}
+
+			++len;
+			HASHSIG_HASH_MIX(state, ch);
+		}
 
-		hashsig_heap_insert(&sig->mins, (hashsig_t)state);
-		hashsig_heap_insert(&sig->maxs, (hashsig_t)state);
-		sig->considered++;
+		if (len > 0) {
+			hashsig_heap_insert(&sig->mins, (hashsig_t)state);
+			hashsig_heap_insert(&sig->maxs, (hashsig_t)state);
 
-		prog->window[prog->win_pos] = ch;
-		prog->win_pos = (prog->win_pos + 1) % HASHSIG_HASH_WINDOW;
+			while (scan < end && (*scan == '\n' || !*scan))
+				++scan;
+		}
 	}
 
-	prog->state = state;
+	prog->use_ignores = use_ignores;
 
 	return 0;
 }
 
 static int hashsig_finalize_hashes(git_hashsig *sig)
 {
-	if (sig->mins.size < HASHSIG_HEAP_SIZE) {
-		giterr_set(GITERR_INVALID,
-			"File too small for similarity signature calculation");
+	if (sig->mins.size < HASHSIG_HEAP_MIN_SIZE &&
+		!(sig->opt & GIT_HASHSIG_ALLOW_SMALL_FILES)) {
+		git_error_set(GIT_ERROR_INVALID,
+			"file too small for similarity signature calculation");
 		return GIT_EBUFS;
 	}
 
@@ -274,11 +247,13 @@ int git_hashsig_create(
 	git_hashsig_option_t opts)
 {
 	int error;
-	hashsig_in_progress prog = HASHSIG_IN_PROGRESS_INIT;
+	hashsig_in_progress prog;
 	git_hashsig *sig = hashsig_alloc(opts);
-	GITERR_CHECK_ALLOC(sig);
+	GIT_ERROR_CHECK_ALLOC(sig);
+
+	hashsig_in_progress_init(&prog, sig);
 
-	error = hashsig_add_hashes(sig, buf, buflen, &prog);
+	error = hashsig_add_hashes(sig, (const uint8_t *)buf, buflen, &prog);
 
 	if (!error)
 		error = hashsig_finalize_hashes(sig);
@@ -296,23 +271,25 @@ int git_hashsig_create_fromfile(
 	const char *path,
 	git_hashsig_option_t opts)
 {
-	char buf[4096];
+	uint8_t buf[0x1000];
 	ssize_t buflen = 0;
 	int error = 0, fd;
-	hashsig_in_progress prog = HASHSIG_IN_PROGRESS_INIT;
+	hashsig_in_progress prog;
 	git_hashsig *sig = hashsig_alloc(opts);
-	GITERR_CHECK_ALLOC(sig);
+	GIT_ERROR_CHECK_ALLOC(sig);
 
 	if ((fd = git_futils_open_ro(path)) < 0) {
 		git__free(sig);
 		return fd;
 	}
 
+	hashsig_in_progress_init(&prog, sig);
+
 	while (!error) {
 		if ((buflen = p_read(fd, buf, sizeof(buf))) <= 0) {
 			if ((error = (int)buflen) < 0)
-				giterr_set(GITERR_OS,
-					"Read error on '%s' calculating similarity hashes", path);
+				git_error_set(GIT_ERROR_OS,
+					"read error on '%s' calculating similarity hashes", path);
 			break;
 		}
 
@@ -362,6 +339,24 @@ static int hashsig_heap_compare(const hashsig_heap *a, const hashsig_heap *b)
 
 int git_hashsig_compare(const git_hashsig *a, const git_hashsig *b)
 {
-	return (hashsig_heap_compare(&a->mins, &b->mins) +
-			hashsig_heap_compare(&a->maxs, &b->maxs)) / 2;
+	/* 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);
+	else
+		return (hashsig_heap_compare(&a->mins, &b->mins) +
+				hashsig_heap_compare(&a->maxs, &b->maxs)) / 2;
 }

data/vendor/libgit2/src/ident.c

@@ -0,0 +1,129 @@
+/*
+ * Copyright (C) the libgit2 contributors. All rights reserved.
+ *
+ * This file is part of libgit2, distributed under the GNU GPL v2 with
+ * a Linking Exception. For full terms see the included COPYING file.
+ */
+
+#include "common.h"
+
+#include "git2/sys/filter.h"
+#include "filter.h"
+#include "buffer.h"
+#include "buf_text.h"
+
+static int ident_find_id(
+	const char **id_start, const char **id_end, const char *start, size_t len)
+{
+	const char *end = start + len, *found = NULL;
+
+	while (len > 3 && (found = memchr(start, '$', len)) != NULL) {
+		size_t remaining = (size_t)(end - found) - 1;
+		if (remaining < 3)
+			return GIT_ENOTFOUND;
+
+		start = found + 1;
+		len   = remaining;
+
+		if (start[0] == 'I' && start[1] == 'd')
+			break;
+	}
+
+	if (len < 3 || !found)
+		return GIT_ENOTFOUND;
+	*id_start = found;
+
+	if ((found = memchr(start + 2, '$', len - 2)) == NULL)
+		return GIT_ENOTFOUND;
+
+	*id_end = found + 1;
+	return 0;
+}
+
+static int ident_insert_id(
+	git_buf *to, const git_buf *from, const git_filter_source *src)
+{
+	char oid[GIT_OID_HEXSZ+1];
+	const char *id_start, *id_end, *from_end = from->ptr + from->size;
+	size_t need_size;
+
+	/* replace $Id$ with blob id */
+
+	if (!git_filter_source_id(src))
+		return GIT_PASSTHROUGH;
+
+	git_oid_tostr(oid, sizeof(oid), git_filter_source_id(src));
+
+	if (ident_find_id(&id_start, &id_end, from->ptr, from->size) < 0)
+		return GIT_PASSTHROUGH;
+
+	need_size = (size_t)(id_start - from->ptr) +
+		5 /* "$Id: " */ + GIT_OID_HEXSZ + 2 /* " $" */ +
+		(size_t)(from_end - id_end);
+
+	if (git_buf_grow(to, need_size) < 0)
+		return -1;
+
+	git_buf_set(to, from->ptr, (size_t)(id_start - from->ptr));
+	git_buf_put(to, "$Id: ", 5);
+	git_buf_put(to, oid, GIT_OID_HEXSZ);
+	git_buf_put(to, " $", 2);
+	git_buf_put(to, id_end, (size_t)(from_end - id_end));
+
+	return git_buf_oom(to) ? -1 : 0;
+}
+
+static int ident_remove_id(
+	git_buf *to, const git_buf *from)
+{
+	const char *id_start, *id_end, *from_end = from->ptr + from->size;
+	size_t need_size;
+
+	if (ident_find_id(&id_start, &id_end, from->ptr, from->size) < 0)
+		return GIT_PASSTHROUGH;
+
+	need_size = (size_t)(id_start - from->ptr) +
+		4 /* "$Id$" */ + (size_t)(from_end - id_end);
+
+	if (git_buf_grow(to, need_size) < 0)
+		return -1;
+
+	git_buf_set(to, from->ptr, (size_t)(id_start - from->ptr));
+	git_buf_put(to, "$Id$", 4);
+	git_buf_put(to, id_end, (size_t)(from_end - id_end));
+
+	return git_buf_oom(to) ? -1 : 0;
+}
+
+static int ident_apply(
+	git_filter     *self,
+	void          **payload,
+	git_buf        *to,
+	const git_buf  *from,
+	const git_filter_source *src)
+{
+	GIT_UNUSED(self); GIT_UNUSED(payload);
+
+	/* Don't filter binary files */
+	if (git_buf_text_is_binary(from))
+		return GIT_PASSTHROUGH;
+
+	if (git_filter_source_mode(src) == GIT_FILTER_SMUDGE)
+		return ident_insert_id(to, from, src);
+	else
+		return ident_remove_id(to, from);
+}
+
+git_filter *git_ident_filter_new(void)
+{
+	git_filter *f = git__calloc(1, sizeof(git_filter));
+	if (f == NULL)
+		return NULL;
+
+	f->version = GIT_FILTER_VERSION;
+	f->attributes = "+ident"; /* apply to files with ident attribute set */
+	f->shutdown = git_filter_free;
+	f->apply    = ident_apply;
+
+	return f;
+}

data/vendor/libgit2/src/idxmap.c

@@ -0,0 +1,153 @@
+/*
+ * Copyright (C) the libgit2 contributors. All rights reserved.
+ *
+ * This file is part of libgit2, distributed under the GNU GPL v2 with
+ * a Linking Exception. For full terms see the included COPYING file.
+ */
+
+#include "idxmap.h"
+
+#define kmalloc git__malloc
+#define kcalloc git__calloc
+#define krealloc git__realloc
+#define kreallocarray git__reallocarray
+#define kfree git__free
+#include "khash.h"
+
+__KHASH_TYPE(idx, const git_index_entry *, git_index_entry *)
+__KHASH_TYPE(idxicase, const git_index_entry *, git_index_entry *)
+
+/* This is __ac_X31_hash_string but with tolower and it takes the entry's stage into account */
+static kh_inline khint_t idxentry_hash(const git_index_entry *e)
+{
+	const char *s = e->path;
+	khint_t h = (khint_t)git__tolower(*s);
+	if (h) for (++s ; *s; ++s) h = (h << 5) - h + (khint_t)git__tolower(*s);
+	return h + GIT_INDEX_ENTRY_STAGE(e);
+}
+
+#define idxentry_equal(a, b) (GIT_INDEX_ENTRY_STAGE(a) == GIT_INDEX_ENTRY_STAGE(b) && strcmp(a->path, b->path) == 0)
+#define idxentry_icase_equal(a, b) (GIT_INDEX_ENTRY_STAGE(a) == GIT_INDEX_ENTRY_STAGE(b) && strcasecmp(a->path, b->path) == 0)
+
+__KHASH_IMPL(idx, static kh_inline, const git_index_entry *, git_index_entry *, 1, idxentry_hash, idxentry_equal)
+__KHASH_IMPL(idxicase, static kh_inline, const git_index_entry *, git_index_entry *, 1, idxentry_hash, idxentry_icase_equal)
+
+int git_idxmap_alloc(git_idxmap **map)
+{
+	if ((*map = kh_init(idx)) == NULL) {
+		git_error_set_oom();
+		return -1;
+	}
+
+	return 0;
+}
+
+int git_idxmap_icase_alloc(git_idxmap_icase **map)
+{
+	if ((*map = kh_init(idxicase)) == NULL) {
+		git_error_set_oom();
+		return -1;
+	}
+
+	return 0;
+}
+
+void git_idxmap_insert(git_idxmap *map, const git_index_entry *key, void *value, int *rval)
+{
+	khiter_t idx = kh_put(idx, map, key, rval);
+
+	if ((*rval) >= 0) {
+		if ((*rval) == 0)
+			kh_key(map, idx) = key;
+		kh_val(map, idx) = value;
+	}
+}
+
+void git_idxmap_icase_insert(git_idxmap_icase *map, const git_index_entry *key, void *value, int *rval)
+{
+	khiter_t idx = kh_put(idxicase, map, key, rval);
+
+	if ((*rval) >= 0) {
+		if ((*rval) == 0)
+			kh_key(map, idx) = key;
+		kh_val(map, idx) = value;
+	}
+}
+
+size_t git_idxmap_lookup_index(git_idxmap *map, const git_index_entry *key)
+{
+	return kh_get(idx, map, key);
+}
+
+size_t git_idxmap_icase_lookup_index(git_idxmap_icase *map, const git_index_entry *key)
+{
+	return kh_get(idxicase, map, key);
+}
+
+void *git_idxmap_value_at(git_idxmap *map, size_t idx)
+{
+	return kh_val(map, idx);
+}
+
+int git_idxmap_valid_index(git_idxmap *map, size_t idx)
+{
+	return idx != kh_end(map);
+}
+
+int git_idxmap_has_data(git_idxmap *map, size_t idx)
+{
+	return kh_exist(map, idx);
+}
+
+void git_idxmap_resize(git_idxmap *map, size_t size)
+{
+	kh_resize(idx, map, size);
+}
+
+void git_idxmap_icase_resize(git_idxmap_icase *map, size_t size)
+{
+	kh_resize(idxicase, map, size);
+}
+
+void git_idxmap_free(git_idxmap *map)
+{
+	kh_destroy(idx, map);
+}
+
+void git_idxmap_icase_free(git_idxmap_icase *map)
+{
+	kh_destroy(idxicase, map);
+}
+
+void git_idxmap_clear(git_idxmap *map)
+{
+	kh_clear(idx, map);
+}
+
+void git_idxmap_icase_clear(git_idxmap_icase *map)
+{
+	kh_clear(idxicase, map);
+}
+
+void git_idxmap_delete_at(git_idxmap *map, size_t idx)
+{
+	kh_del(idx, map, idx);
+}
+
+void git_idxmap_icase_delete_at(git_idxmap_icase *map, size_t idx)
+{
+	kh_del(idxicase, map, idx);
+}
+
+void git_idxmap_delete(git_idxmap *map, const git_index_entry *key)
+{
+	khiter_t idx = git_idxmap_lookup_index(map, key);
+	if (git_idxmap_valid_index(map, idx))
+		git_idxmap_delete_at(map, idx);
+}
+void git_idxmap_icase_delete(git_idxmap_icase *map, const git_index_entry *key)
+{
+	khiter_t idx = git_idxmap_icase_lookup_index(map, key);
+	if (git_idxmap_valid_index((git_idxmap *)map, idx))
+		git_idxmap_icase_delete_at(map, idx);
+}

data/vendor/libgit2/src/idxmap.h

@@ -0,0 +1,41 @@
+/*
+ * Copyright (C) the libgit2 contributors. All rights reserved.
+ *
+ * This file is part of libgit2, distributed under the GNU GPL v2 with
+ * a Linking Exception. For full terms see the included COPYING file.
+ */
+#ifndef INCLUDE_idxmap_h__
+#define INCLUDE_idxmap_h__
+
+#include "common.h"
+
+#include "git2/index.h"
+
+typedef struct kh_idx_s git_idxmap;
+typedef struct kh_idxicase_s git_idxmap_icase;
+
+int git_idxmap_alloc(git_idxmap **map);
+int git_idxmap_icase_alloc(git_idxmap_icase **map);
+void git_idxmap_insert(git_idxmap *map, const git_index_entry *key, void *value, int *rval);
+void git_idxmap_icase_insert(git_idxmap_icase *map, const git_index_entry *key, void *value, int *rval);
+
+size_t git_idxmap_lookup_index(git_idxmap *map, const git_index_entry *key);
+size_t git_idxmap_icase_lookup_index(git_idxmap_icase *map, const git_index_entry *key);
+void *git_idxmap_value_at(git_idxmap *map, size_t idx);
+int git_idxmap_valid_index(git_idxmap *map, size_t idx);
+int git_idxmap_has_data(git_idxmap *map, size_t idx);
+
+void git_idxmap_resize(git_idxmap *map, size_t size);
+void git_idxmap_icase_resize(git_idxmap_icase *map, size_t size);
+void git_idxmap_free(git_idxmap *map);
+void git_idxmap_icase_free(git_idxmap_icase *map);
+void git_idxmap_clear(git_idxmap *map);
+void git_idxmap_icase_clear(git_idxmap_icase *map);
+
+void git_idxmap_delete_at(git_idxmap *map, size_t idx);
+void git_idxmap_icase_delete_at(git_idxmap_icase *map, size_t idx);
+
+void git_idxmap_delete(git_idxmap *map, const git_index_entry *key);
+void git_idxmap_icase_delete(git_idxmap_icase *map, const git_index_entry *key);
+
+#endif