[libgit2.git] / src / diff_file.c

/*
 * 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/blob.h"
#include "git2/submodule.h"
#include "diff.h"
#include "diff_file.h"
#include "odb.h"
#include "fileops.h"
#include "filter.h"

#define DIFF_MAX_FILESIZE 0x20000000

static bool diff_file_content_binary_by_size(git_diff_file_content *fc)
{
	/* if we have diff opts, check max_size vs file size */
	if ((fc->file->flags & DIFF_FLAGS_KNOWN_BINARY) == 0 &&
		fc->opts_max_size > 0 &&
		fc->file->size > fc->opts_max_size)
		fc->file->flags |= GIT_DIFF_FLAG_BINARY;

	return ((fc->file->flags & GIT_DIFF_FLAG_BINARY) != 0);
}

static void diff_file_content_binary_by_content(git_diff_file_content *fc)
{
	if ((fc->file->flags & DIFF_FLAGS_KNOWN_BINARY) != 0)
		return;

	switch (git_diff_driver_content_is_binary(
		fc->driver, fc->map.data, fc->map.len)) {
	case 0: fc->file->flags |= GIT_DIFF_FLAG_NOT_BINARY; break;
	case 1: fc->file->flags |= GIT_DIFF_FLAG_BINARY; break;
	default: break;
	}
}

static int diff_file_content_init_common(
	git_diff_file_content *fc, const git_diff_options *opts)
{
	fc->opts_flags = opts ? opts->flags : GIT_DIFF_NORMAL;

	if (opts && opts->max_size >= 0)
		fc->opts_max_size = opts->max_size ?
			opts->max_size : DIFF_MAX_FILESIZE;

	if (fc->src == GIT_ITERATOR_TYPE_EMPTY)
		fc->src = GIT_ITERATOR_TYPE_TREE;

	if (!fc->driver &&
		git_diff_driver_lookup(&fc->driver, fc->repo, fc->file->path) < 0)
		return -1;

	/* give driver a chance to modify options */
	git_diff_driver_update_options(&fc->opts_flags, fc->driver);

	/* make sure file is conceivable mmap-able */
	if ((git_off_t)((size_t)fc->file->size) != fc->file->size)
		fc->file->flags |= GIT_DIFF_FLAG_BINARY;
	/* check if user is forcing text diff the file */
	else if (fc->opts_flags & GIT_DIFF_FORCE_TEXT) {
		fc->file->flags &= ~GIT_DIFF_FLAG_BINARY;
		fc->file->flags |= GIT_DIFF_FLAG_NOT_BINARY;
	}
	/* check if user is forcing binary diff the file */
	else if (fc->opts_flags & GIT_DIFF_FORCE_BINARY) {
		fc->file->flags &= ~GIT_DIFF_FLAG_NOT_BINARY;
		fc->file->flags |= GIT_DIFF_FLAG_BINARY;
	}

	diff_file_content_binary_by_size(fc);

	if ((fc->flags & GIT_DIFF_FLAG__NO_DATA) != 0) {
		fc->flags |= GIT_DIFF_FLAG__LOADED;
		fc->map.len  = 0;
		fc->map.data = "";
	}

	if ((fc->flags & GIT_DIFF_FLAG__LOADED) != 0)
		diff_file_content_binary_by_content(fc);

	return 0;
}

int git_diff_file_content__init_from_diff(
	git_diff_file_content *fc,
	git_diff *diff,
	size_t delta_index,
	bool use_old)
{
	git_diff_delta *delta = git_vector_get(&diff->deltas, delta_index);
	bool has_data = true;

	memset(fc, 0, sizeof(*fc));
	fc->repo = diff->repo;
	fc->file = use_old ? &delta->old_file : &delta->new_file;
	fc->src  = use_old ? diff->old_src : diff->new_src;

	if (git_diff_driver_lookup(&fc->driver, fc->repo, fc->file->path) < 0)
		return -1;

	switch (delta->status) {
	case GIT_DELTA_ADDED:
		has_data = !use_old; break;
	case GIT_DELTA_DELETED:
		has_data = use_old; break;
	case GIT_DELTA_UNTRACKED:
		has_data = !use_old &&
			(diff->opts.flags & GIT_DIFF_SHOW_UNTRACKED_CONTENT) != 0;
		break;
	case GIT_DELTA_UNREADABLE:
	case GIT_DELTA_MODIFIED:
	case GIT_DELTA_COPIED:
	case GIT_DELTA_RENAMED:
		break;
	default:
		has_data = false;
		break;
	}

	if (!has_data)
		fc->flags |= GIT_DIFF_FLAG__NO_DATA;

	return diff_file_content_init_common(fc, &diff->opts);
}

int git_diff_file_content__init_from_src(
	git_diff_file_content *fc,
	git_repository *repo,
	const git_diff_options *opts,
	const git_diff_file_content_src *src,
	git_diff_file *as_file)
{
	memset(fc, 0, sizeof(*fc));
	fc->repo = repo;
	fc->file = as_file;
	fc->blob = src->blob;

	if (!src->blob && !src->buf) {
		fc->flags |= GIT_DIFF_FLAG__NO_DATA;
	} else {
		fc->flags |= GIT_DIFF_FLAG__LOADED;
		fc->file->flags |= GIT_DIFF_FLAG_VALID_ID;
		fc->file->mode = GIT_FILEMODE_BLOB;

		if (src->blob) {
			fc->file->size = git_blob_rawsize(src->blob);
			git_oid_cpy(&fc->file->id, git_blob_id(src->blob));

			fc->map.len  = (size_t)fc->file->size;
			fc->map.data = (char *)git_blob_rawcontent(src->blob);
		} else {
			fc->file->size = src->buflen;
			git_odb_hash(&fc->file->id, src->buf, src->buflen, GIT_OBJ_BLOB);

			fc->map.len  = src->buflen;
			fc->map.data = (char *)src->buf;
		}
	}

	return diff_file_content_init_common(fc, opts);
}

static int diff_file_content_commit_to_str(
	git_diff_file_content *fc, bool check_status)
{
	char oid[GIT_OID_HEXSZ+1];
	git_buf content = GIT_BUF_INIT;
	const char *status = "";

	if (check_status) {
		int error = 0;
		git_submodule *sm = NULL;
		unsigned int sm_status = 0;
		const git_oid *sm_head;

		if ((error = git_submodule_lookup(&sm, fc->repo, fc->file->path)) < 0) {
			/* GIT_EEXISTS means a "submodule" that has not been git added */
			if (error == GIT_EEXISTS) {
				giterr_clear();
				error = 0;
			}
			return error;
		}

		if ((error = git_submodule_status(&sm_status, sm)) < 0) {
			git_submodule_free(sm);
			return error;
		}

		/* update OID if we didn't have it previously */
		if ((fc->file->flags & GIT_DIFF_FLAG_VALID_ID) == 0 &&
			((sm_head = git_submodule_wd_id(sm)) != NULL ||
			 (sm_head = git_submodule_head_id(sm)) != NULL))
		{
			git_oid_cpy(&fc->file->id, sm_head);
			fc->file->flags |= GIT_DIFF_FLAG_VALID_ID;
		}

		if (GIT_SUBMODULE_STATUS_IS_WD_DIRTY(sm_status))
			status = "-dirty";

		git_submodule_free(sm);
	}

	git_oid_tostr(oid, sizeof(oid), &fc->file->id);
	if (git_buf_printf(&content, "Subproject commit %s%s\n", oid, status) < 0)
		return -1;

	fc->map.len  = git_buf_len(&content);
	fc->map.data = git_buf_detach(&content);
	fc->flags |= GIT_DIFF_FLAG__FREE_DATA;

	return 0;
}

static int diff_file_content_load_blob(git_diff_file_content *fc)
{
	int error = 0;
	git_odb_object *odb_obj = NULL;

	if (git_oid_iszero(&fc->file->id))
		return 0;

	if (fc->file->mode == GIT_FILEMODE_COMMIT)
		return diff_file_content_commit_to_str(fc, false);

	/* if we don't know size, try to peek at object header first */
	if (!fc->file->size) {
		if ((error = git_diff_file__resolve_zero_size(
				fc->file, &odb_obj, fc->repo)) < 0)
			return error;
	}

	if (diff_file_content_binary_by_size(fc))
		return 0;

	if (odb_obj != NULL) {
		error = git_object__from_odb_object(
			(git_object **)&fc->blob, fc->repo, odb_obj, GIT_OBJ_BLOB);
		git_odb_object_free(odb_obj);
	} else {
		error = git_blob_lookup(
			(git_blob **)&fc->blob, fc->repo, &fc->file->id);
	}

	if (!error) {
		fc->flags |= GIT_DIFF_FLAG__FREE_BLOB;
		fc->map.data = (void *)git_blob_rawcontent(fc->blob);
		fc->map.len  = (size_t)git_blob_rawsize(fc->blob);
	}

	return error;
}

static int diff_file_content_load_workdir_symlink(
	git_diff_file_content *fc, git_buf *path)
{
	ssize_t alloc_len, read_len;

	/* link path on disk could be UTF-16, so prepare a buffer that is
	 * big enough to handle some UTF-8 data expansion
	 */
	alloc_len = (ssize_t)(fc->file->size * 2) + 1;

	fc->map.data = git__calloc(alloc_len, sizeof(char));
	GITERR_CHECK_ALLOC(fc->map.data);

	fc->flags |= GIT_DIFF_FLAG__FREE_DATA;

	read_len = p_readlink(git_buf_cstr(path), fc->map.data, alloc_len);
	if (read_len < 0) {
		giterr_set(GITERR_OS, "Failed to read symlink '%s'", fc->file->path);
		return -1;
	}

	fc->map.len = read_len;
	return 0;
}

static int diff_file_content_load_workdir_file(
	git_diff_file_content *fc, git_buf *path)
{
	int error = 0;
	git_filter_list *fl = NULL;
	git_file fd = git_futils_open_ro(git_buf_cstr(path));
	git_buf raw = GIT_BUF_INIT;

	if (fd < 0)
		return fd;

	if (!fc->file->size &&
		!(fc->file->size = git_futils_filesize(fd)))
		goto cleanup;

	if (diff_file_content_binary_by_size(fc))
		goto cleanup;

	if ((error = git_filter_list_load(
			&fl, fc->repo, NULL, fc->file->path,
			GIT_FILTER_TO_ODB, GIT_FILTER_OPT_ALLOW_UNSAFE)) < 0)
		goto cleanup;

	/* if there are no filters, try to mmap the file */
	if (fl == NULL) {
		if (!(error = git_futils_mmap_ro(
				&fc->map, fd, 0, (size_t)fc->file->size))) {
			fc->flags |= GIT_DIFF_FLAG__UNMAP_DATA;
			goto cleanup;
		}

		/* if mmap failed, fall through to try readbuffer below */
		giterr_clear();
	}

	if (!(error = git_futils_readbuffer_fd(&raw, fd, (size_t)fc->file->size))) {
		git_buf out = GIT_BUF_INIT;

		error = git_filter_list_apply_to_data(&out, fl, &raw);

		if (out.ptr != raw.ptr)
			git_buf_free(&raw);

		if (!error) {
			fc->map.len  = out.size;
			fc->map.data = out.ptr;
			fc->flags |= GIT_DIFF_FLAG__FREE_DATA;
		}
	}

cleanup:
	git_filter_list_free(fl);
	p_close(fd);

	return error;
}

static int diff_file_content_load_workdir(git_diff_file_content *fc)
{
	int error = 0;
	git_buf path = GIT_BUF_INIT;

	if (fc->file->mode == GIT_FILEMODE_COMMIT)
		return diff_file_content_commit_to_str(fc, true);

	if (fc->file->mode == GIT_FILEMODE_TREE)
		return 0;

	if (git_buf_joinpath(
			&path, git_repository_workdir(fc->repo), fc->file->path) < 0)
		return -1;

	if (S_ISLNK(fc->file->mode))
		error = diff_file_content_load_workdir_symlink(fc, &path);
	else
		error = diff_file_content_load_workdir_file(fc, &path);

	/* once data is loaded, update OID if we didn't have it previously */
	if (!error && (fc->file->flags & GIT_DIFF_FLAG_VALID_ID) == 0) {
		error = git_odb_hash(
			&fc->file->id, fc->map.data, fc->map.len, GIT_OBJ_BLOB);
		fc->file->flags |= GIT_DIFF_FLAG_VALID_ID;
	}

	git_buf_free(&path);
	return error;
}

int git_diff_file_content__load(git_diff_file_content *fc)
{
	int error = 0;

	if ((fc->flags & GIT_DIFF_FLAG__LOADED) != 0)
		return 0;

	if ((fc->file->flags & GIT_DIFF_FLAG_BINARY) != 0)
		return 0;

	if (fc->src == GIT_ITERATOR_TYPE_WORKDIR)
		error = diff_file_content_load_workdir(fc);
	else
		error = diff_file_content_load_blob(fc);
	if (error)
		return error;

	fc->flags |= GIT_DIFF_FLAG__LOADED;

	diff_file_content_binary_by_content(fc);

	return 0;
}

void git_diff_file_content__unload(git_diff_file_content *fc)
{
	if ((fc->flags & GIT_DIFF_FLAG__LOADED) == 0)
		return;

	if (fc->flags & GIT_DIFF_FLAG__FREE_DATA) {
		git__free(fc->map.data);
		fc->map.data = "";
		fc->map.len  = 0;
		fc->flags &= ~GIT_DIFF_FLAG__FREE_DATA;
	}
	else if (fc->flags & GIT_DIFF_FLAG__UNMAP_DATA) {
		git_futils_mmap_free(&fc->map);
		fc->map.data = "";
		fc->map.len  = 0;
		fc->flags &= ~GIT_DIFF_FLAG__UNMAP_DATA;
	}

	if (fc->flags & GIT_DIFF_FLAG__FREE_BLOB) {
		git_blob_free((git_blob *)fc->blob);
		fc->blob = NULL;
		fc->flags &= ~GIT_DIFF_FLAG__FREE_BLOB;
	}

	fc->flags &= ~GIT_DIFF_FLAG__LOADED;
}

void git_diff_file_content__clear(git_diff_file_content *fc)
{
	git_diff_file_content__unload(fc);

	/* for now, nothing else to do */
}
Commit	Line	Data
114f5a6c RB	1	/*
	2	* Copyright (C) the libgit2 contributors. All rights reserved.
	3	*
	4	* This file is part of libgit2, distributed under the GNU GPL v2 with
	5	* a Linking Exception. For full terms see the included COPYING file.
	6	*/
	7	#include "common.h"
	8	#include "git2/blob.h"
	9	#include "git2/submodule.h"
	10	#include "diff.h"
	11	#include "diff_file.h"
	12	#include "odb.h"
	13	#include "fileops.h"
	14	#include "filter.h"
	15
	16	#define DIFF_MAX_FILESIZE 0x20000000
	17
	18	static bool diff_file_content_binary_by_size(git_diff_file_content *fc)
	19	{
	20	/* if we have diff opts, check max_size vs file size */
74ded024	21	if ((fc->file->flags & DIFF_FLAGS_KNOWN_BINARY) == 0 &&
5dc98298	22	fc->opts_max_size > 0 &&
74ded024 RB	23	fc->file->size > fc->opts_max_size)
74ded024 RB	24	fc->file->flags \|= GIT_DIFF_FLAG_BINARY;
114f5a6c	25
74ded024	26	return ((fc->file->flags & GIT_DIFF_FLAG_BINARY) != 0);
114f5a6c RB	27	}
	28
	29	static void diff_file_content_binary_by_content(git_diff_file_content *fc)
	30	{
74ded024	31	if ((fc->file->flags & DIFF_FLAGS_KNOWN_BINARY) != 0)
114f5a6c RB	32	return;
	33
	34	switch (git_diff_driver_content_is_binary(
	35	fc->driver, fc->map.data, fc->map.len)) {
74ded024 RB	36	case 0: fc->file->flags \|= GIT_DIFF_FLAG_NOT_BINARY; break;
74ded024 RB	37	case 1: fc->file->flags \|= GIT_DIFF_FLAG_BINARY; break;
114f5a6c RB	38	default: break;
	39	}
	40	}
	41
5dc98298 RB	42	static int diff_file_content_init_common(
5dc98298 RB	43	git_diff_file_content fc, const git_diff_options opts)
114f5a6c	44	{
5dc98298 RB	45	fc->opts_flags = opts ? opts->flags : GIT_DIFF_NORMAL;
	46
	47	if (opts && opts->max_size >= 0)
	48	fc->opts_max_size = opts->max_size ?
	49	opts->max_size : DIFF_MAX_FILESIZE;
114f5a6c	50
74ded024	51	if (fc->src == GIT_ITERATOR_TYPE_EMPTY)
114f5a6c	52	fc->src = GIT_ITERATOR_TYPE_TREE;
74ded024 RB	53
	54	if (!fc->driver &&
	55	git_diff_driver_lookup(&fc->driver, fc->repo, fc->file->path) < 0)
	56	return -1;
114f5a6c	57
5dc98298 RB	58	/* give driver a chance to modify options */
	59	git_diff_driver_update_options(&fc->opts_flags, fc->driver);
	60
114f5a6c	61	/* make sure file is conceivable mmap-able */
74ded024 RB	62	if ((git_off_t)((size_t)fc->file->size) != fc->file->size)
74ded024 RB	63	fc->file->flags \|= GIT_DIFF_FLAG_BINARY;
5dc98298 RB	64	/* check if user is forcing text diff the file */
5dc98298 RB	65	else if (fc->opts_flags & GIT_DIFF_FORCE_TEXT) {
74ded024 RB	66	fc->file->flags &= ~GIT_DIFF_FLAG_BINARY;
74ded024 RB	67	fc->file->flags \|= GIT_DIFF_FLAG_NOT_BINARY;
5dc98298 RB	68	}
	69	/* check if user is forcing binary diff the file */
	70	else if (fc->opts_flags & GIT_DIFF_FORCE_BINARY) {
74ded024 RB	71	fc->file->flags &= ~GIT_DIFF_FLAG_NOT_BINARY;
74ded024 RB	72	fc->file->flags \|= GIT_DIFF_FLAG_BINARY;
5dc98298	73	}
114f5a6c RB	74
	75	diff_file_content_binary_by_size(fc);
	76
74ded024 RB	77	if ((fc->flags & GIT_DIFF_FLAG__NO_DATA) != 0) {
74ded024 RB	78	fc->flags \|= GIT_DIFF_FLAG__LOADED;
114f5a6c RB	79	fc->map.len = 0;
	80	fc->map.data = "";
	81	}
	82
74ded024	83	if ((fc->flags & GIT_DIFF_FLAG__LOADED) != 0)
114f5a6c RB	84	diff_file_content_binary_by_content(fc);
	85
	86	return 0;
	87	}
	88
360f42f4	89	int git_diff_file_content__init_from_diff(
114f5a6c	90	git_diff_file_content *fc,
3ff1d123	91	git_diff *diff,
114f5a6c RB	92	size_t delta_index,
	93	bool use_old)
	94	{
	95	git_diff_delta *delta = git_vector_get(&diff->deltas, delta_index);
114f5a6c RB	96	bool has_data = true;
	97
	98	memset(fc, 0, sizeof(*fc));
	99	fc->repo = diff->repo;
74ded024	100	fc->file = use_old ? &delta->old_file : &delta->new_file;
114f5a6c	101	fc->src = use_old ? diff->old_src : diff->new_src;
114f5a6c	102
74ded024	103	if (git_diff_driver_lookup(&fc->driver, fc->repo, fc->file->path) < 0)
114f5a6c RB	104	return -1;
	105
	106	switch (delta->status) {
	107	case GIT_DELTA_ADDED:
	108	has_data = !use_old; break;
	109	case GIT_DELTA_DELETED:
	110	has_data = use_old; break;
	111	case GIT_DELTA_UNTRACKED:
	112	has_data = !use_old &&
10672e3e	113	(diff->opts.flags & GIT_DIFF_SHOW_UNTRACKED_CONTENT) != 0;
114f5a6c	114	break;
61bef72d	115	case GIT_DELTA_UNREADABLE:
114f5a6c RB	116	case GIT_DELTA_MODIFIED:
	117	case GIT_DELTA_COPIED:
	118	case GIT_DELTA_RENAMED:
	119	break;
	120	default:
	121	has_data = false;
	122	break;
	123	}
	124
	125	if (!has_data)
74ded024	126	fc->flags \|= GIT_DIFF_FLAG__NO_DATA;
114f5a6c	127
5dc98298	128	return diff_file_content_init_common(fc, &diff->opts);
114f5a6c RB	129	}
114f5a6c RB	130
6789b7a7	131	int git_diff_file_content__init_from_src(
114f5a6c RB	132	git_diff_file_content *fc,
	133	git_repository *repo,
	134	const git_diff_options *opts,
6789b7a7	135	const git_diff_file_content_src *src,
74ded024	136	git_diff_file *as_file)
114f5a6c RB	137	{
	138	memset(fc, 0, sizeof(*fc));
	139	fc->repo = repo;
74ded024	140	fc->file = as_file;
6789b7a7	141	fc->blob = src->blob;
114f5a6c	142
6789b7a7	143	if (!src->blob && !src->buf) {
74ded024	144	fc->flags \|= GIT_DIFF_FLAG__NO_DATA;
114f5a6c	145	} else {
74ded024	146	fc->flags \|= GIT_DIFF_FLAG__LOADED;
9950bb4e	147	fc->file->flags \|= GIT_DIFF_FLAG_VALID_ID;
74ded024	148	fc->file->mode = GIT_FILEMODE_BLOB;
114f5a6c	149
6789b7a7 RB	150	if (src->blob) {
	151	fc->file->size = git_blob_rawsize(src->blob);
	152	git_oid_cpy(&fc->file->id, git_blob_id(src->blob));
114f5a6c	153
6789b7a7 RB	154	fc->map.len = (size_t)fc->file->size;
	155	fc->map.data = (char *)git_blob_rawcontent(src->blob);
	156	} else {
	157	fc->file->size = src->buflen;
	158	git_odb_hash(&fc->file->id, src->buf, src->buflen, GIT_OBJ_BLOB);
114f5a6c	159
6789b7a7 RB	160	fc->map.len = src->buflen;
	161	fc->map.data = (char *)src->buf;
	162	}
114f5a6c RB	163	}
114f5a6c RB	164
5dc98298	165	return diff_file_content_init_common(fc, opts);
114f5a6c RB	166	}
	167
	168	static int diff_file_content_commit_to_str(
	169	git_diff_file_content *fc, bool check_status)
	170	{
	171	char oid[GIT_OID_HEXSZ+1];
	172	git_buf content = GIT_BUF_INIT;
	173	const char *status = "";
	174
	175	if (check_status) {
	176	int error = 0;
	177	git_submodule *sm = NULL;
	178	unsigned int sm_status = 0;
	179	const git_oid *sm_head;
	180
a15c7802	181	if ((error = git_submodule_lookup(&sm, fc->repo, fc->file->path)) < 0) {
114f5a6c	182	/* GIT_EEXISTS means a "submodule" that has not been git added */
a15c7802 RB	183	if (error == GIT_EEXISTS) {
a15c7802 RB	184	giterr_clear();
114f5a6c	185	error = 0;
a15c7802 RB	186	}
	187	return error;
	188	}
	189
	190	if ((error = git_submodule_status(&sm_status, sm)) < 0) {
	191	git_submodule_free(sm);
114f5a6c RB	192	return error;
	193	}
	194
	195	/* update OID if we didn't have it previously */
9950bb4e	196	if ((fc->file->flags & GIT_DIFF_FLAG_VALID_ID) == 0 &&
114f5a6c RB	197	((sm_head = git_submodule_wd_id(sm)) != NULL \|\|
	198	(sm_head = git_submodule_head_id(sm)) != NULL))
	199	{
9950bb4e CMN	200	git_oid_cpy(&fc->file->id, sm_head);
9950bb4e CMN	201	fc->file->flags \|= GIT_DIFF_FLAG_VALID_ID;
114f5a6c RB	202	}
	203
	204	if (GIT_SUBMODULE_STATUS_IS_WD_DIRTY(sm_status))
	205	status = "-dirty";
a15c7802 RB	206
a15c7802 RB	207	git_submodule_free(sm);
114f5a6c RB	208	}
114f5a6c RB	209
9950bb4e	210	git_oid_tostr(oid, sizeof(oid), &fc->file->id);
114f5a6c RB	211	if (git_buf_printf(&content, "Subproject commit %s%s\n", oid, status) < 0)
	212	return -1;
	213
	214	fc->map.len = git_buf_len(&content);
	215	fc->map.data = git_buf_detach(&content);
74ded024	216	fc->flags \|= GIT_DIFF_FLAG__FREE_DATA;
114f5a6c RB	217
	218	return 0;
	219	}
	220
	221	static int diff_file_content_load_blob(git_diff_file_content *fc)
	222	{
	223	int error = 0;
	224	git_odb_object *odb_obj = NULL;
	225
9950bb4e	226	if (git_oid_iszero(&fc->file->id))
114f5a6c RB	227	return 0;
114f5a6c RB	228
74ded024	229	if (fc->file->mode == GIT_FILEMODE_COMMIT)
114f5a6c RB	230	return diff_file_content_commit_to_str(fc, false);
	231
	232	/* if we don't know size, try to peek at object header first */
74ded024	233	if (!fc->file->size) {
effdbeb3 RB	234	if ((error = git_diff_file__resolve_zero_size(
effdbeb3 RB	235	fc->file, &odb_obj, fc->repo)) < 0)
114f5a6c	236	return error;
114f5a6c RB	237	}
	238
	239	if (diff_file_content_binary_by_size(fc))
	240	return 0;
	241
	242	if (odb_obj != NULL) {
	243	error = git_object__from_odb_object(
	244	(git_object **)&fc->blob, fc->repo, odb_obj, GIT_OBJ_BLOB);
	245	git_odb_object_free(odb_obj);
	246	} else {
	247	error = git_blob_lookup(
9950bb4e	248	(git_blob **)&fc->blob, fc->repo, &fc->file->id);
114f5a6c RB	249	}
	250
	251	if (!error) {
74ded024	252	fc->flags \|= GIT_DIFF_FLAG__FREE_BLOB;
114f5a6c RB	253	fc->map.data = (void *)git_blob_rawcontent(fc->blob);
	254	fc->map.len = (size_t)git_blob_rawsize(fc->blob);
	255	}
	256
	257	return error;
	258	}
	259
	260	static int diff_file_content_load_workdir_symlink(
	261	git_diff_file_content fc, git_buf path)
	262	{
	263	ssize_t alloc_len, read_len;
	264
	265	/* link path on disk could be UTF-16, so prepare a buffer that is
	266	* big enough to handle some UTF-8 data expansion
	267	*/
74ded024	268	alloc_len = (ssize_t)(fc->file->size * 2) + 1;
114f5a6c RB	269
	270	fc->map.data = git__calloc(alloc_len, sizeof(char));
	271	GITERR_CHECK_ALLOC(fc->map.data);
	272
74ded024	273	fc->flags \|= GIT_DIFF_FLAG__FREE_DATA;
114f5a6c RB	274
	275	read_len = p_readlink(git_buf_cstr(path), fc->map.data, alloc_len);
	276	if (read_len < 0) {
74ded024	277	giterr_set(GITERR_OS, "Failed to read symlink '%s'", fc->file->path);
114f5a6c RB	278	return -1;
	279	}
	280
	281	fc->map.len = read_len;
	282	return 0;
	283	}
	284
	285	static int diff_file_content_load_workdir_file(
	286	git_diff_file_content fc, git_buf path)
	287	{
	288	int error = 0;
85d54812	289	git_filter_list *fl = NULL;
114f5a6c	290	git_file fd = git_futils_open_ro(git_buf_cstr(path));
2a7d224f	291	git_buf raw = GIT_BUF_INIT;
114f5a6c RB	292
	293	if (fd < 0)
	294	return fd;
	295
74ded024 RB	296	if (!fc->file->size &&
74ded024 RB	297	!(fc->file->size = git_futils_filesize(fd)))
114f5a6c RB	298	goto cleanup;
	299
	300	if (diff_file_content_binary_by_size(fc))
	301	goto cleanup;
	302
85d54812	303	if ((error = git_filter_list_load(
5269008c RB	304	&fl, fc->repo, NULL, fc->file->path,
5269008c RB	305	GIT_FILTER_TO_ODB, GIT_FILTER_OPT_ALLOW_UNSAFE)) < 0)
114f5a6c	306	goto cleanup;
114f5a6c	307
85d54812 RB	308	/* if there are no filters, try to mmap the file */
85d54812 RB	309	if (fl == NULL) {
114f5a6c	310	if (!(error = git_futils_mmap_ro(
85d54812	311	&fc->map, fd, 0, (size_t)fc->file->size))) {
74ded024	312	fc->flags \|= GIT_DIFF_FLAG__UNMAP_DATA;
114f5a6c	313	goto cleanup;
114f5a6c RB	314	}
114f5a6c RB	315
85d54812 RB	316	/* if mmap failed, fall through to try readbuffer below */
85d54812 RB	317	giterr_clear();
114f5a6c RB	318	}
114f5a6c RB	319
2a7d224f	320	if (!(error = git_futils_readbuffer_fd(&raw, fd, (size_t)fc->file->size))) {
a9f51e43	321	git_buf out = GIT_BUF_INIT;
2a7d224f	322
a9f51e43	323	error = git_filter_list_apply_to_data(&out, fl, &raw);
2a7d224f	324
591e8295 RB	325	if (out.ptr != raw.ptr)
591e8295 RB	326	git_buf_free(&raw);
85d54812	327
2a7d224f RB	328	if (!error) {
	329	fc->map.len = out.size;
	330	fc->map.data = out.ptr;
	331	fc->flags \|= GIT_DIFF_FLAG__FREE_DATA;
	332	}
	333	}
85d54812	334
114f5a6c	335	cleanup:
85d54812	336	git_filter_list_free(fl);
114f5a6c RB	337	p_close(fd);
	338
	339	return error;
	340	}
	341
	342	static int diff_file_content_load_workdir(git_diff_file_content *fc)
	343	{
	344	int error = 0;
	345	git_buf path = GIT_BUF_INIT;
	346
74ded024	347	if (fc->file->mode == GIT_FILEMODE_COMMIT)
114f5a6c RB	348	return diff_file_content_commit_to_str(fc, true);
114f5a6c RB	349
74ded024	350	if (fc->file->mode == GIT_FILEMODE_TREE)
114f5a6c RB	351	return 0;
	352
	353	if (git_buf_joinpath(
74ded024	354	&path, git_repository_workdir(fc->repo), fc->file->path) < 0)
114f5a6c RB	355	return -1;
114f5a6c RB	356
74ded024	357	if (S_ISLNK(fc->file->mode))
114f5a6c RB	358	error = diff_file_content_load_workdir_symlink(fc, &path);
	359	else
	360	error = diff_file_content_load_workdir_file(fc, &path);
	361
	362	/* once data is loaded, update OID if we didn't have it previously */
9950bb4e	363	if (!error && (fc->file->flags & GIT_DIFF_FLAG_VALID_ID) == 0) {
114f5a6c	364	error = git_odb_hash(
9950bb4e CMN	365	&fc->file->id, fc->map.data, fc->map.len, GIT_OBJ_BLOB);
9950bb4e CMN	366	fc->file->flags \|= GIT_DIFF_FLAG_VALID_ID;
114f5a6c RB	367	}
	368
	369	git_buf_free(&path);
	370	return error;
	371	}
	372
360f42f4	373	int git_diff_file_content__load(git_diff_file_content *fc)
114f5a6c RB	374	{
	375	int error = 0;
	376
74ded024	377	if ((fc->flags & GIT_DIFF_FLAG__LOADED) != 0)
114f5a6c RB	378	return 0;
114f5a6c RB	379
74ded024	380	if ((fc->file->flags & GIT_DIFF_FLAG_BINARY) != 0)
114f5a6c RB	381	return 0;
	382
	383	if (fc->src == GIT_ITERATOR_TYPE_WORKDIR)
	384	error = diff_file_content_load_workdir(fc);
	385	else
	386	error = diff_file_content_load_blob(fc);
	387	if (error)
	388	return error;
	389
74ded024	390	fc->flags \|= GIT_DIFF_FLAG__LOADED;
114f5a6c RB	391
	392	diff_file_content_binary_by_content(fc);
	393
	394	return 0;
	395	}
	396
360f42f4	397	void git_diff_file_content__unload(git_diff_file_content *fc)
114f5a6c	398	{
39a1a662 RB	399	if ((fc->flags & GIT_DIFF_FLAG__LOADED) == 0)
	400	return;
	401
74ded024	402	if (fc->flags & GIT_DIFF_FLAG__FREE_DATA) {
114f5a6c RB	403	git__free(fc->map.data);
	404	fc->map.data = "";
	405	fc->map.len = 0;
74ded024	406	fc->flags &= ~GIT_DIFF_FLAG__FREE_DATA;
114f5a6c	407	}
74ded024	408	else if (fc->flags & GIT_DIFF_FLAG__UNMAP_DATA) {
114f5a6c RB	409	git_futils_mmap_free(&fc->map);
	410	fc->map.data = "";
	411	fc->map.len = 0;
74ded024	412	fc->flags &= ~GIT_DIFF_FLAG__UNMAP_DATA;
114f5a6c RB	413	}
114f5a6c RB	414
74ded024	415	if (fc->flags & GIT_DIFF_FLAG__FREE_BLOB) {
114f5a6c RB	416	git_blob_free((git_blob *)fc->blob);
114f5a6c RB	417	fc->blob = NULL;
74ded024	418	fc->flags &= ~GIT_DIFF_FLAG__FREE_BLOB;
114f5a6c RB	419	}
114f5a6c RB	420
74ded024	421	fc->flags &= ~GIT_DIFF_FLAG__LOADED;
114f5a6c RB	422	}
114f5a6c RB	423
360f42f4	424	void git_diff_file_content__clear(git_diff_file_content *fc)
114f5a6c	425	{
360f42f4	426	git_diff_file_content__unload(fc);
114f5a6c RB	427
	428	/* for now, nothing else to do */
	429	}