kmx git

/*
 * 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/config.h"
#include "git2/blob.h"

#include "diff.h"
#include "hashsig.h"
#include "path.h"
#include "fileops.h"

static git_diff_delta *diff_delta__dup(
	const git_diff_delta *d, git_pool *pool)
{
	git_diff_delta *delta = git__malloc(sizeof(git_diff_delta));
	if (!delta)
		return NULL;

	memcpy(delta, d, sizeof(git_diff_delta));
	GIT_DIFF_FLAG__CLEAR_INTERNAL(delta->flags);

	if (d->old_file.path != NULL) {
		delta->old_file.path = git_pool_strdup(pool, d->old_file.path);
		if (delta->old_file.path == NULL)
			goto fail;
	}

	if (d->new_file.path != d->old_file.path && d->new_file.path != NULL) {
		delta->new_file.path = git_pool_strdup(pool, d->new_file.path);
		if (delta->new_file.path == NULL)
			goto fail;
	} else {
		delta->new_file.path = delta->old_file.path;
	}

	return delta;

fail:
	git__free(delta);
	return NULL;
}

static git_diff_delta *diff_delta__merge_like_cgit(
	const git_diff_delta *a, const git_diff_delta *b, git_pool *pool)
{
	git_diff_delta *dup;

	/* Emulate C git for merging two diffs (a la 'git diff <sha>').
	 *
	 * When C git does a diff between the work dir and a tree, it actually
	 * diffs with the index but uses the workdir contents.  This emulates
	 * those choices so we can emulate the type of diff.
	 *
	 * We have three file descriptions here, let's call them:
	 *  f1 = a->old_file
	 *  f2 = a->new_file AND b->old_file
	 *  f3 = b->new_file
	 */

	/* if f2 == f3 or f2 is deleted, then just dup the 'a' diff */
	if (b->status == GIT_DELTA_UNMODIFIED || a->status == GIT_DELTA_DELETED)
		return diff_delta__dup(a, pool);

	/* otherwise, base this diff on the 'b' diff */
	if ((dup = diff_delta__dup(b, pool)) == NULL)
		return NULL;

	/* If 'a' status is uninteresting, then we're done */
	if (a->status == GIT_DELTA_UNMODIFIED)
		return dup;

	assert(a->status != GIT_DELTA_UNMODIFIED);
	assert(b->status != GIT_DELTA_UNMODIFIED);

	/* A cgit exception is that the diff of a file that is only in the
	 * index (i.e. not in HEAD nor workdir) is given as empty.
	 */
	if (dup->status == GIT_DELTA_DELETED) {
		if (a->status == GIT_DELTA_ADDED)
			dup->status = GIT_DELTA_UNMODIFIED;
		/* else don't overwrite DELETE status */
	} else {
		dup->status = a->status;
	}

	git_oid_cpy(&dup->old_file.oid, &a->old_file.oid);
	dup->old_file.mode  = a->old_file.mode;
	dup->old_file.size  = a->old_file.size;
	dup->old_file.flags = a->old_file.flags;

	return dup;
}

int git_diff_merge(
	git_diff_list *onto,
	const git_diff_list *from)
{
	int error = 0;
	git_pool onto_pool;
	git_vector onto_new;
	git_diff_delta *delta;
	bool ignore_case = false;
	unsigned int i, j;

	assert(onto && from);

	if (!from->deltas.length)
		return 0;

	if (git_vector_init(
			&onto_new, onto->deltas.length, git_diff_delta__cmp) < 0 ||
		git_pool_init(&onto_pool, 1, 0) < 0)
		return -1;

	if ((onto->opts.flags & GIT_DIFF_DELTAS_ARE_ICASE) != 0 ||
		(from->opts.flags & GIT_DIFF_DELTAS_ARE_ICASE) != 0)
	{
		ignore_case = true;

		/* This function currently only supports merging diff lists that
		 * are sorted identically. */
		assert((onto->opts.flags & GIT_DIFF_DELTAS_ARE_ICASE) != 0 &&
				(from->opts.flags & GIT_DIFF_DELTAS_ARE_ICASE) != 0);
	}

	for (i = 0, j = 0; i < onto->deltas.length || j < from->deltas.length; ) {
		git_diff_delta *o = GIT_VECTOR_GET(&onto->deltas, i);
		const git_diff_delta *f = GIT_VECTOR_GET(&from->deltas, j);
		int cmp = !f ? -1 : !o ? 1 : STRCMP_CASESELECT(ignore_case, o->old_file.path, f->old_file.path);

		if (cmp < 0) {
			delta = diff_delta__dup(o, &onto_pool);
			i++;
		} else if (cmp > 0) {
			delta = diff_delta__dup(f, &onto_pool);
			j++;
		} else {
			delta = diff_delta__merge_like_cgit(o, f, &onto_pool);
			i++;
			j++;
		}

		/* the ignore rules for the target may not match the source
		 * or the result of a merged delta could be skippable...
		 */
		if (git_diff_delta__should_skip(&onto->opts, delta)) {
			git__free(delta);
			continue;
		}

		if ((error = !delta ? -1 : git_vector_insert(&onto_new, delta)) < 0)
			break;
	}

	if (!error) {
		git_vector_swap(&onto->deltas, &onto_new);
		git_pool_swap(&onto->pool, &onto_pool);
		onto->new_src = from->new_src;

		/* prefix strings also come from old pool, so recreate those.*/
		onto->opts.old_prefix =
			git_pool_strdup_safe(&onto->pool, onto->opts.old_prefix);
		onto->opts.new_prefix =
			git_pool_strdup_safe(&onto->pool, onto->opts.new_prefix);
	}

	git_vector_foreach(&onto_new, i, delta)
		git__free(delta);
	git_vector_free(&onto_new);
	git_pool_clear(&onto_pool);

	return error;
}

int git_diff_find_similar__hashsig_for_file(
	void **out, const git_diff_file *f, const char *path, void *p)
{
	git_hashsig_option_t opt = (git_hashsig_option_t)p;
	int error = 0;

	GIT_UNUSED(f);
	error = git_hashsig_create_fromfile((git_hashsig **)out, path, opt);

	if (error == GIT_EBUFS) {
		error = 0;
		giterr_clear();
	}

	return error;
}

int git_diff_find_similar__hashsig_for_buf(
	void **out, const git_diff_file *f, const char *buf, size_t len, void *p)
{
	git_hashsig_option_t opt = (git_hashsig_option_t)p;
	int error = 0;

	GIT_UNUSED(f);
	error = git_hashsig_create((git_hashsig **)out, buf, len, opt);

	if (error == GIT_EBUFS) {
		error = 0;
		giterr_clear();
	}

	return error;
}

void git_diff_find_similar__hashsig_free(void *sig, void *payload)
{
	GIT_UNUSED(payload);
	git_hashsig_free(sig);
}

int git_diff_find_similar__calc_similarity(
	int *score, void *siga, void *sigb, void *payload)
{
	GIT_UNUSED(payload);
	*score = git_hashsig_compare(siga, sigb);
	return 0;
}

#define DEFAULT_THRESHOLD 50
#define DEFAULT_BREAK_REWRITE_THRESHOLD 60
#define DEFAULT_RENAME_LIMIT 200

static int normalize_find_opts(
	git_diff_list *diff,
	git_diff_find_options *opts,
	git_diff_find_options *given)
{
	git_config *cfg = NULL;

	if (diff->repo != NULL &&
		git_repository_config__weakptr(&cfg, diff->repo) < 0)
		return -1;

	if (given != NULL)
		memcpy(opts, given, sizeof(*opts));
	else {
		const char *val = NULL;

		GIT_INIT_STRUCTURE(opts, GIT_DIFF_FIND_OPTIONS_VERSION);

		opts->flags = GIT_DIFF_FIND_RENAMES;

		if (git_config_get_string(&val, cfg, "diff.renames") < 0)
			giterr_clear();
		else if (val &&
			(!strcasecmp(val, "copies") || !strcasecmp(val, "copy")))
			opts->flags = GIT_DIFF_FIND_RENAMES | GIT_DIFF_FIND_COPIES;
	}

	GITERR_CHECK_VERSION(opts, GIT_DIFF_FIND_OPTIONS_VERSION, "git_diff_find_options");

	/* some flags imply others */

	if (opts->flags & GIT_DIFF_FIND_EXACT_MATCH_ONLY) {
		/* if we are only looking for exact matches, then don't turn
		 * MODIFIED items into ADD/DELETE pairs because it's too picky
		 */
		opts->flags &= ~(GIT_DIFF_FIND_REWRITES | GIT_DIFF_BREAK_REWRITES);

		/* similarly, don't look for self-rewrites to split */
		opts->flags &= ~GIT_DIFF_FIND_RENAMES_FROM_REWRITES;
	}

	if (opts->flags & GIT_DIFF_FIND_RENAMES_FROM_REWRITES)
		opts->flags |= GIT_DIFF_FIND_RENAMES;

	if (opts->flags & GIT_DIFF_FIND_COPIES_FROM_UNMODIFIED)
		opts->flags |= GIT_DIFF_FIND_COPIES;

	if (opts->flags & GIT_DIFF_BREAK_REWRITES)
		opts->flags |= GIT_DIFF_FIND_REWRITES;

#define USE_DEFAULT(X) ((X) == 0 || (X) > 100)

	if (USE_DEFAULT(opts->rename_threshold))
		opts->rename_threshold = DEFAULT_THRESHOLD;

	if (USE_DEFAULT(opts->rename_from_rewrite_threshold))
		opts->rename_from_rewrite_threshold = DEFAULT_THRESHOLD;

	if (USE_DEFAULT(opts->copy_threshold))
		opts->copy_threshold = DEFAULT_THRESHOLD;

	if (USE_DEFAULT(opts->break_rewrite_threshold))
		opts->break_rewrite_threshold = DEFAULT_BREAK_REWRITE_THRESHOLD;

#undef USE_DEFAULT

	if (!opts->rename_limit) {
		int32_t limit = 0;

		opts->rename_limit = DEFAULT_RENAME_LIMIT;

		if (git_config_get_int32(&limit, cfg, "diff.renameLimit") < 0)
			giterr_clear();
		else if (limit > 0)
			opts->rename_limit = limit;
	}

	/* assign the internal metric with whitespace flag as payload */
	if (!opts->metric) {
		opts->metric = git__malloc(sizeof(git_diff_similarity_metric));
		GITERR_CHECK_ALLOC(opts->metric);

		opts->metric->file_signature = git_diff_find_similar__hashsig_for_file;
		opts->metric->buffer_signature = git_diff_find_similar__hashsig_for_buf;
		opts->metric->free_signature = git_diff_find_similar__hashsig_free;
		opts->metric->similarity = git_diff_find_similar__calc_similarity;

		if (opts->flags & GIT_DIFF_FIND_IGNORE_WHITESPACE)
			opts->metric->payload = (void *)GIT_HASHSIG_IGNORE_WHITESPACE;
		else if (opts->flags & GIT_DIFF_FIND_DONT_IGNORE_WHITESPACE)
			opts->metric->payload = (void *)GIT_HASHSIG_NORMAL;
		else
			opts->metric->payload = (void *)GIT_HASHSIG_SMART_WHITESPACE;
	}

	return 0;
}

static int apply_splits_and_deletes(
	git_diff_list *diff, size_t expected_size, bool actually_split)
{
	git_vector onto = GIT_VECTOR_INIT;
	size_t i;
	git_diff_delta *delta, *deleted;

	if (git_vector_init(&onto, expected_size, git_diff_delta__cmp) < 0)
		return -1;

	/* build new delta list without TO_DELETE and splitting TO_SPLIT */
	git_vector_foreach(&diff->deltas, i, delta) {
		if ((delta->flags & GIT_DIFF_FLAG__TO_DELETE) != 0)
			continue;

		if ((delta->flags & GIT_DIFF_FLAG__TO_SPLIT) != 0 && actually_split) {
			delta->similarity = 0;

			/* make new record for DELETED side of split */
			if (!(deleted = diff_delta__dup(delta, &diff->pool)))
				goto on_error;

			deleted->status = GIT_DELTA_DELETED;
			memset(&deleted->new_file, 0, sizeof(deleted->new_file));
			deleted->new_file.path = deleted->old_file.path;
			deleted->new_file.flags |= GIT_DIFF_FLAG_VALID_OID;

			if (git_vector_insert(&onto, deleted) < 0)
				goto on_error;

			if (diff->new_src == GIT_ITERATOR_TYPE_WORKDIR)
				delta->status = GIT_DELTA_UNTRACKED;
			else
				delta->status = GIT_DELTA_ADDED;
			memset(&delta->old_file, 0, sizeof(delta->old_file));
			delta->old_file.path = delta->new_file.path;
			delta->old_file.flags |= GIT_DIFF_FLAG_VALID_OID;
		}

		/* clean up delta before inserting into new list */
		GIT_DIFF_FLAG__CLEAR_INTERNAL(delta->flags);

		if (delta->status != GIT_DELTA_COPIED &&
			delta->status != GIT_DELTA_RENAMED &&
			(delta->status != GIT_DELTA_MODIFIED || actually_split))
			delta->similarity = 0;

		/* insert into new list */
		if (git_vector_insert(&onto, delta) < 0)
			goto on_error;
	}

	/* cannot return an error past this point */

	/* free deltas from old list that didn't make it to the new one */
	git_vector_foreach(&diff->deltas, i, delta) {
		if ((delta->flags & GIT_DIFF_FLAG__TO_DELETE) != 0)
			git__free(delta);
	}

	/* swap new delta list into place */
	git_vector_swap(&diff->deltas, &onto);
	git_vector_free(&onto);
	git_vector_sort(&diff->deltas);

	return 0;

on_error:
	git_vector_foreach(&onto, i, delta)
		git__free(delta);
	git_vector_free(&onto);

	return -1;
}

GIT_INLINE(git_diff_file *) similarity_get_file(git_diff_list *diff, size_t idx)
{
	git_diff_delta *delta = git_vector_get(&diff->deltas, idx / 2);
	return (idx & 1) ? &delta->new_file : &delta->old_file;
}

typedef struct {
	size_t idx;
	git_iterator_type_t src;
	git_repository *repo;
	git_diff_file *file;
	git_buf data;
	git_odb_object *odb_obj;
	git_blob *blob;
} similarity_info;

static int similarity_init(
	similarity_info *info, git_diff_list *diff, size_t file_idx)
{
	info->idx  = file_idx;
	info->src  = (file_idx & 1) ? diff->new_src : diff->old_src;
	info->repo = diff->repo;
	info->file = similarity_get_file(diff, file_idx);
	info->odb_obj = NULL;
	info->blob = NULL;
	git_buf_init(&info->data, 0);

	if (info->file->size > 0)
		return 0;

	return git_diff_file__resolve_zero_size(
		info->file, &info->odb_obj, info->repo);
}

static int similarity_calc(
	similarity_info *info,
	const git_diff_find_options *opts,
	void **cache)
{
	int error = 0;
	git_diff_file *file = info->file;

	if (info->src == GIT_ITERATOR_TYPE_WORKDIR) {
		if ((error = git_buf_joinpath(
			&info->data, git_repository_workdir(info->repo), file->path)) < 0)
			return error;

		/* if path is not a regular file, just skip this item */
		if (!git_path_isfile(info->data.ptr))
			return 0;

		/* TODO: apply wd-to-odb filters to file data if necessary */

		error = opts->metric->file_signature(
			&cache[info->idx], info->file,
			info->data.ptr, opts->metric->payload);
	} else {
		/* if we didn't initially know the size, we might have an odb_obj
		 * around from earlier, so convert that, otherwise load the blob now
		 */
		if (info->odb_obj != NULL)
			error = git_object__from_odb_object(
				(git_object **)&info->blob, info->repo,
				info->odb_obj, GIT_OBJ_BLOB);
		else
			error = git_blob_lookup(&info->blob, info->repo, &file->oid);

		if (error < 0) {
			/* if lookup fails, just skip this item in similarity calc */
			giterr_clear();
		} else {
			size_t sz;

			/* index size may not be actual blob size if filtered */
			if (file->size != git_blob_rawsize(info->blob))
				file->size = git_blob_rawsize(info->blob);

			sz = (size_t)(git__is_sizet(file->size) ? file->size : -1);

			error = opts->metric->buffer_signature(
				&cache[info->idx], info->file,
				git_blob_rawcontent(info->blob), sz, opts->metric->payload);
		}
	}

	return error;
}

static void similarity_unload(similarity_info *info)
{
	if (info->odb_obj)
		git_odb_object_free(info->odb_obj);

	if (info->blob)
		git_blob_free(info->blob);
	else
		git_buf_free(&info->data);
}

#define FLAG_SET(opts,flag_name) (((opts)->flags & flag_name) != 0)

/* - score < 0 means files cannot be compared
 * - score >= 100 means files are exact match
 * - score == 0 means files are completely different
 */
static int similarity_measure(
	int *score,
	git_diff_list *diff,
	const git_diff_find_options *opts,
	void **cache,
	size_t a_idx,
	size_t b_idx)
{
	git_diff_file *a_file = similarity_get_file(diff, a_idx);
	git_diff_file *b_file = similarity_get_file(diff, b_idx);
	bool exact_match = FLAG_SET(opts, GIT_DIFF_FIND_EXACT_MATCH_ONLY);
	int error = 0;
	similarity_info a_info, b_info;

	*score = -1;

	/* don't try to compare files of different types */
	if (GIT_MODE_TYPE(a_file->mode) != GIT_MODE_TYPE(b_file->mode))
		return 0;

	/* if exact match is requested, force calculation of missing OIDs now */
	if (exact_match) {
		if (git_oid_iszero(&a_file->oid) &&
			diff->old_src == GIT_ITERATOR_TYPE_WORKDIR &&
			!git_diff__oid_for_file(diff->repo, a_file->path,
				a_file->mode, a_file->size, &a_file->oid))
			a_file->flags |= GIT_DIFF_FLAG_VALID_OID;

		if (git_oid_iszero(&b_file->oid) &&
			diff->new_src == GIT_ITERATOR_TYPE_WORKDIR &&
			!git_diff__oid_for_file(diff->repo, b_file->path,
				b_file->mode, b_file->size, &b_file->oid))
			b_file->flags |= GIT_DIFF_FLAG_VALID_OID;
	}

	/* check OID match as a quick test */
	if (git_oid__cmp(&a_file->oid, &b_file->oid) == 0) {
		*score = 100;
		return 0;
	}

	/* don't calculate signatures if we are doing exact match */
	if (exact_match) {
		*score = 0;
		return 0;
	}

	memset(&a_info, 0, sizeof(a_info));
	memset(&b_info, 0, sizeof(b_info));

	/* set up similarity data (will try to update missing file sizes) */
	if (!cache[a_idx] && (error = similarity_init(&a_info, diff, a_idx)) < 0)
		return error;
	if (!cache[b_idx] && (error = similarity_init(&b_info, diff, b_idx)) < 0)
		goto cleanup;

	/* check if file sizes are nowhere near each other */
	if (a_file->size > 127 &&
		b_file->size > 127 &&
		(a_file->size > (b_file->size << 4) ||
		 b_file->size > (a_file->size << 4)))
		goto cleanup;

	/* update signature cache if needed */
	if (!cache[a_idx] && (error = similarity_calc(&a_info, opts, cache)) < 0)
		goto cleanup;

	if (!cache[b_idx] && (error = similarity_calc(&b_info, opts, cache)) < 0)
		goto cleanup;

	/* calculate similarity provided that the metric choose to process
	 * both the a and b files (some may not if file is too big, etc).
	 */
	if (cache[a_idx] && cache[b_idx])
		error = opts->metric->similarity(
			score, cache[a_idx], cache[b_idx], opts->metric->payload);

cleanup:
	similarity_unload(&a_info);
	similarity_unload(&b_info);

	return error;
}

static int calc_self_similarity(
	git_diff_list *diff,
	const git_diff_find_options *opts,
	size_t delta_idx,
	void **cache)
{
	int error, similarity = -1;
	git_diff_delta *delta = GIT_VECTOR_GET(&diff->deltas, delta_idx);

	if ((delta->flags & GIT_DIFF_FLAG__HAS_SELF_SIMILARITY) != 0)
		return 0;

	error = similarity_measure(
		&similarity, diff, opts, cache, 2 * delta_idx, 2 * delta_idx + 1);
	if (error < 0)
		return error;

	if (similarity >= 0) {
		delta->similarity = (uint32_t)similarity;
		delta->flags |= GIT_DIFF_FLAG__HAS_SELF_SIMILARITY;
	}

	return 0;
}

static bool is_rename_target(
	git_diff_list *diff,
	const git_diff_find_options *opts,
	size_t delta_idx,
	void **cache)
{
	git_diff_delta *delta = GIT_VECTOR_GET(&diff->deltas, delta_idx);

	/* skip things that aren't plain blobs */
	if (!GIT_MODE_ISBLOB(delta->new_file.mode))
		return false;

	/* only consider ADDED, RENAMED, COPIED, and split MODIFIED as
	 * targets; maybe include UNTRACKED and IGNORED if requested.
	 */
	switch (delta->status) {
	case GIT_DELTA_UNMODIFIED:
	case GIT_DELTA_DELETED:
		return false;

	case GIT_DELTA_MODIFIED:
		if (!FLAG_SET(opts, GIT_DIFF_FIND_REWRITES) &&
			!FLAG_SET(opts, GIT_DIFF_FIND_RENAMES_FROM_REWRITES))
			return false;

		if (calc_self_similarity(diff, opts, delta_idx, cache) < 0)
			return false;

		if (FLAG_SET(opts, GIT_DIFF_BREAK_REWRITES) &&
			delta->similarity < opts->break_rewrite_threshold) {
			delta->flags |= GIT_DIFF_FLAG__TO_SPLIT;
			break;
		}
		if (FLAG_SET(opts, GIT_DIFF_FIND_RENAMES_FROM_REWRITES) &&
			delta->similarity < opts->rename_from_rewrite_threshold)
			break;

		return false;

	case GIT_DELTA_UNTRACKED:
		if (!FLAG_SET(opts, GIT_DIFF_FIND_FOR_UNTRACKED))
			return false;
		break;

	case GIT_DELTA_IGNORED:
		return false;

	default: /* all other status values should be checked */
		break;
	}

	delta->flags |= GIT_DIFF_FLAG__IS_RENAME_TARGET;
	return true;
}

static bool is_rename_source(
	git_diff_list *diff,
	const git_diff_find_options *opts,
	size_t delta_idx,
	void **cache)
{
	git_diff_delta *delta = GIT_VECTOR_GET(&diff->deltas, delta_idx);

	/* skip things that aren't blobs */
	if (!GIT_MODE_ISBLOB(delta->old_file.mode))
		return false;

	switch (delta->status) {
	case GIT_DELTA_ADDED:
	case GIT_DELTA_UNTRACKED:
	case GIT_DELTA_IGNORED:
		return false;

	case GIT_DELTA_DELETED:
	case GIT_DELTA_TYPECHANGE:
		break;

	case GIT_DELTA_UNMODIFIED:
		if (!FLAG_SET(opts, GIT_DIFF_FIND_COPIES_FROM_UNMODIFIED))
			return false;
		break;

	default: /* MODIFIED, RENAMED, COPIED */
		/* if we're finding copies, this could be a source */
		if (FLAG_SET(opts, GIT_DIFF_FIND_COPIES))
			break;

		/* otherwise, this is only a source if we can split it */
		if (!FLAG_SET(opts, GIT_DIFF_FIND_REWRITES) &&
			!FLAG_SET(opts, GIT_DIFF_FIND_RENAMES_FROM_REWRITES))
			return false;

		if (calc_self_similarity(diff, opts, delta_idx, cache) < 0)
			return false;

		if (FLAG_SET(opts, GIT_DIFF_BREAK_REWRITES) &&
			delta->similarity < opts->break_rewrite_threshold) {
			delta->flags |= GIT_DIFF_FLAG__TO_SPLIT;
			break;
		}

		if (FLAG_SET(opts, GIT_DIFF_FIND_RENAMES_FROM_REWRITES) &&
			delta->similarity < opts->rename_from_rewrite_threshold)
			break;

		return false;
	}

	delta->flags |= GIT_DIFF_FLAG__IS_RENAME_SOURCE;
	return true;
}

GIT_INLINE(bool) delta_is_split(git_diff_delta *delta)
{
	return (delta->status == GIT_DELTA_TYPECHANGE ||
			(delta->flags & GIT_DIFF_FLAG__TO_SPLIT) != 0);
}

GIT_INLINE(bool) delta_is_new_only(git_diff_delta *delta)
{
	return (delta->status == GIT_DELTA_ADDED ||
			delta->status == GIT_DELTA_UNTRACKED ||
			delta->status == GIT_DELTA_IGNORED);
}

GIT_INLINE(void) delta_make_rename(
	git_diff_delta *to, const git_diff_delta *from, uint32_t similarity)
{
	to->status     = GIT_DELTA_RENAMED;
	to->similarity = similarity;
	memcpy(&to->old_file, &from->old_file, sizeof(to->old_file));
	to->flags &= ~GIT_DIFF_FLAG__TO_SPLIT;
}

typedef struct {
	uint32_t idx;
	uint32_t similarity;
} diff_find_match;

int git_diff_find_similar(
	git_diff_list *diff,
	git_diff_find_options *given_opts)
{
	size_t i, j, sigcache_size;
	int error = 0, similarity;
	git_diff_delta *from, *to;
	git_diff_find_options opts;
	size_t num_srcs = 0, num_tgts = 0, tried_srcs = 0, tried_tgts = 0;
	size_t num_rewrites = 0, num_updates = 0, num_bumped = 0;
	void **sigcache; /* cache of similarity metric file signatures */
	diff_find_match *match_srcs = NULL, *match_tgts = NULL, *best_match;
	git_diff_file swap;

	if ((error = normalize_find_opts(diff, &opts, given_opts)) < 0)
		return error;

	/* TODO: maybe abort if deltas.length > rename_limit ??? */
	if (!git__is_uint32(diff->deltas.length))
		return 0;

	sigcache_size = diff->deltas.length * 2; /* keep size b/c diff may change */
	sigcache = git__calloc(sigcache_size, sizeof(void *));
	GITERR_CHECK_ALLOC(sigcache);

	/* Label rename sources and targets
	 *
	 * This will also set self-similarity scores for MODIFIED files and
	 * mark them for splitting if break-rewrites is enabled
	 */
	git_vector_foreach(&diff->deltas, i, to) {
		if (is_rename_source(diff, &opts, i, sigcache))
			++num_srcs;

		if (is_rename_target(diff, &opts, i, sigcache))
			++num_tgts;
	}

	/* if there are no candidate srcs or tgts, we're done */
	if (!num_srcs || !num_tgts)
		goto cleanup;

	match_tgts = git__calloc(diff->deltas.length, sizeof(diff_find_match));
	GITERR_CHECK_ALLOC(match_tgts);
	match_srcs = git__calloc(diff->deltas.length, sizeof(diff_find_match));
	GITERR_CHECK_ALLOC(match_srcs);

	/*
	 * Find best-fit matches for rename / copy candidates
	 */

find_best_matches:
	tried_tgts = num_bumped = 0;

	git_vector_foreach(&diff->deltas, i, to) {
		/* skip things that are not rename targets */
		if ((to->flags & GIT_DIFF_FLAG__IS_RENAME_TARGET) == 0)
			continue;

		tried_srcs = 0;

		git_vector_foreach(&diff->deltas, j, from) {
			/* skip things that are not rename sources */
			if ((from->flags & GIT_DIFF_FLAG__IS_RENAME_SOURCE) == 0)
				continue;

			/* calculate similarity for this pair and find best match */
			if (i == j)
				similarity = -1; /* don't measure self-similarity here */
			else if ((error = similarity_measure(
				&similarity, diff, &opts, sigcache, 2 * j, 2 * i + 1)) < 0)
				goto cleanup;

			/* if this pairing is better for the src and the tgt, keep it */
			if (similarity > 0 &&
				match_tgts[i].similarity < (uint32_t)similarity &&
				match_srcs[j].similarity < (uint32_t)similarity)
			{
				if (match_tgts[i].similarity > 0) {
					match_tgts[match_srcs[j].idx].similarity = 0;
					match_srcs[match_tgts[i].idx].similarity = 0;
					++num_bumped;
				}

				match_tgts[i].similarity = (uint32_t)similarity;
				match_tgts[i].idx = (uint32_t)j;

				match_srcs[j].similarity = (uint32_t)similarity;
				match_srcs[j].idx = (uint32_t)i;
			}

			if (++tried_srcs >= num_srcs)
				break;

			/* cap on maximum targets we'll examine (per "to" file) */
			if (tried_srcs > opts.rename_limit)
				break;
		}

		if (++tried_tgts >= num_tgts)
			break;
	}

	if (num_bumped > 0) /* try again if we bumped some items */
		goto find_best_matches;

	/*
	 * Rewrite the diffs with renames / copies
	 */

	tried_tgts = 0;

	git_vector_foreach(&diff->deltas, i, to) {
		/* skip things that are not rename targets */
		if ((to->flags & GIT_DIFF_FLAG__IS_RENAME_TARGET) == 0)
			continue;

		/* check if this delta was the target of a similarity */
		best_match = &match_tgts[i];
		if (!best_match->similarity)
			continue;

		j = best_match->idx;
		from = GIT_VECTOR_GET(&diff->deltas, j);

		/* possible scenarios:
		 * 1. from DELETE to ADD/UNTRACK/IGNORE = RENAME
		 * 2. from DELETE to SPLIT/TYPECHANGE = RENAME + DELETE
		 * 3. from SPLIT/TYPECHANGE to ADD/UNTRACK/IGNORE = ADD + RENAME
		 * 4. from SPLIT/TYPECHANGE to SPLIT/TYPECHANGE = RENAME + SPLIT
		 * 5. from OTHER to ADD/UNTRACK/IGNORE = OTHER + COPY
		 */

		if (from->status == GIT_DELTA_DELETED) {

			if (delta_is_new_only(to)) {

				if (best_match->similarity < opts.rename_threshold)
					continue;

				delta_make_rename(to, from, best_match->similarity);

				from->flags |= GIT_DIFF_FLAG__TO_DELETE;
				num_rewrites++;
			} else {
				assert(delta_is_split(to));

				if (best_match->similarity < opts.rename_from_rewrite_threshold)
					continue;

				memcpy(&swap, &to->old_file, sizeof(swap));

				delta_make_rename(to, from, best_match->similarity);
				num_rewrites--;

				from->status = GIT_DELTA_DELETED;
				memcpy(&from->old_file, &swap, sizeof(from->old_file));
				memset(&from->new_file, 0, sizeof(from->new_file));
				from->new_file.path = from->old_file.path;
				from->new_file.flags |= GIT_DIFF_FLAG_VALID_OID;

				num_updates++;
			}
		}

		else if (delta_is_split(from)) {

			if (delta_is_new_only(to)) {

				if (best_match->similarity < opts.rename_threshold)
					continue;

				delta_make_rename(to, from, best_match->similarity);

				from->status = (diff->new_src == GIT_ITERATOR_TYPE_WORKDIR) ?
					GIT_DELTA_UNTRACKED : GIT_DELTA_ADDED;
				memset(&from->old_file, 0, sizeof(from->old_file));
				from->old_file.path = from->new_file.path;
				from->old_file.flags |= GIT_DIFF_FLAG_VALID_OID;

				from->flags &= ~GIT_DIFF_FLAG__TO_SPLIT;
				num_rewrites--;

				num_updates++;
			} else {
				assert(delta_is_split(from));

				if (best_match->similarity < opts.rename_from_rewrite_threshold)
					continue;

				memcpy(&swap, &to->old_file, sizeof(swap));

				delta_make_rename(to, from, best_match->similarity);
				num_rewrites--;
				num_updates++;

				memcpy(&from->old_file, &swap, sizeof(from->old_file));

				/* if we've just swapped the new element into the correct
				 * place, clear the SPLIT flag
				 */
				if (match_tgts[j].idx == i &&
					match_tgts[j].similarity >
					opts.rename_from_rewrite_threshold) {

					from->status     = GIT_DELTA_RENAMED;
					from->similarity = match_tgts[j].similarity;
					match_tgts[j].similarity = 0;
					from->flags &= ~GIT_DIFF_FLAG__TO_SPLIT;
					num_rewrites--;
				}
				/* otherwise, if we just overwrote a source, update mapping */
				else if (j > i && match_srcs[i].similarity > 0) {
					match_tgts[match_srcs[i].idx].idx = (uint32_t)j;
				}

				num_updates++;
			}
		}

		else if (delta_is_new_only(to)) {
			if (!FLAG_SET(&opts, GIT_DIFF_FIND_COPIES) ||
				best_match->similarity < opts.copy_threshold)
				continue;

			to->status     = GIT_DELTA_COPIED;
			to->similarity = best_match->similarity;
			memcpy(&to->old_file, &from->old_file, sizeof(to->old_file));

			num_updates++;
		}
	}

	/*
	 * Actually split and delete entries as needed
	 */

	if (num_rewrites > 0 || num_updates > 0)
		error = apply_splits_and_deletes(
			diff, diff->deltas.length - num_rewrites,
			FLAG_SET(&opts, GIT_DIFF_BREAK_REWRITES));

cleanup:
	git__free(match_srcs);
	git__free(match_tgts);

	for (i = 0; i < sigcache_size; ++i) {
		if (sigcache[i] != NULL)
			opts.metric->free_signature(sigcache[i], opts.metric->payload);
	}
	git__free(sigcache);

	if (!given_opts || !given_opts->metric)
		git__free(opts.metric);

	return error;
}

#undef FLAG_SET
thodg/libgit2/src/diff_tform.c

Commit

src/diff_tform.c
