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 
8 #include "diff_file.h"
9 
10 #include "git2/blob.h"
11 #include "git2/submodule.h"
12 #include "diff.h"
13 #include "diff_generate.h"
14 #include "odb.h"
15 #include "futils.h"
16 #include "filter.h"
17 
18 #define DIFF_MAX_FILESIZE 0x20000000
19 
diff_file_content_binary_by_size(git_diff_file_content * fc)20 static bool diff_file_content_binary_by_size(git_diff_file_content *fc)
21 {
22 	/* if we have diff opts, check max_size vs file size */
23 	if ((fc->file->flags & DIFF_FLAGS_KNOWN_BINARY) == 0 &&
24 		fc->opts_max_size > 0 &&
25 		fc->file->size > fc->opts_max_size)
26 		fc->file->flags |= GIT_DIFF_FLAG_BINARY;
27 
28 	return ((fc->file->flags & GIT_DIFF_FLAG_BINARY) != 0);
29 }
30 
diff_file_content_binary_by_content(git_diff_file_content * fc)31 static void diff_file_content_binary_by_content(git_diff_file_content *fc)
32 {
33 	if ((fc->file->flags & DIFF_FLAGS_KNOWN_BINARY) != 0)
34 		return;
35 
36 	switch (git_diff_driver_content_is_binary(
37 		fc->driver, fc->map.data, fc->map.len)) {
38 	case 0: fc->file->flags |= GIT_DIFF_FLAG_NOT_BINARY; break;
39 	case 1: fc->file->flags |= GIT_DIFF_FLAG_BINARY; break;
40 	default: break;
41 	}
42 }
43 
diff_file_content_init_common(git_diff_file_content * fc,const git_diff_options * opts)44 static int diff_file_content_init_common(
45 	git_diff_file_content *fc, const git_diff_options *opts)
46 {
47 	fc->opts_flags = opts ? opts->flags : GIT_DIFF_NORMAL;
48 
49 	if (opts && opts->max_size >= 0)
50 		fc->opts_max_size = opts->max_size ?
51 			opts->max_size : DIFF_MAX_FILESIZE;
52 
53 	if (fc->src == GIT_ITERATOR_EMPTY)
54 		fc->src = GIT_ITERATOR_TREE;
55 
56 	if (!fc->driver &&
57 		git_diff_driver_lookup(&fc->driver, fc->repo,
58 		    NULL, fc->file->path) < 0)
59 		return -1;
60 
61 	/* give driver a chance to modify options */
62 	git_diff_driver_update_options(&fc->opts_flags, fc->driver);
63 
64 	/* make sure file is conceivable mmap-able */
65 	if ((size_t)fc->file->size != fc->file->size)
66 		fc->file->flags |= GIT_DIFF_FLAG_BINARY;
67 	/* check if user is forcing text diff the file */
68 	else if (fc->opts_flags & GIT_DIFF_FORCE_TEXT) {
69 		fc->file->flags &= ~GIT_DIFF_FLAG_BINARY;
70 		fc->file->flags |= GIT_DIFF_FLAG_NOT_BINARY;
71 	}
72 	/* check if user is forcing binary diff the file */
73 	else if (fc->opts_flags & GIT_DIFF_FORCE_BINARY) {
74 		fc->file->flags &= ~GIT_DIFF_FLAG_NOT_BINARY;
75 		fc->file->flags |= GIT_DIFF_FLAG_BINARY;
76 	}
77 
78 	diff_file_content_binary_by_size(fc);
79 
80 	if ((fc->flags & GIT_DIFF_FLAG__NO_DATA) != 0) {
81 		fc->flags |= GIT_DIFF_FLAG__LOADED;
82 		fc->map.len  = 0;
83 		fc->map.data = "";
84 	}
85 
86 	if ((fc->flags & GIT_DIFF_FLAG__LOADED) != 0)
87 		diff_file_content_binary_by_content(fc);
88 
89 	return 0;
90 }
91 
git_diff_file_content__init_from_diff(git_diff_file_content * fc,git_diff * diff,git_diff_delta * delta,bool use_old)92 int git_diff_file_content__init_from_diff(
93 	git_diff_file_content *fc,
94 	git_diff *diff,
95 	git_diff_delta *delta,
96 	bool use_old)
97 {
98 	bool has_data = true;
99 
100 	memset(fc, 0, sizeof(*fc));
101 	fc->repo = diff->repo;
102 	fc->file = use_old ? &delta->old_file : &delta->new_file;
103 	fc->src  = use_old ? diff->old_src : diff->new_src;
104 
105 	if (git_diff_driver_lookup(&fc->driver, fc->repo,
106 		    &diff->attrsession, fc->file->path) < 0)
107 		return -1;
108 
109 	switch (delta->status) {
110 	case GIT_DELTA_ADDED:
111 		has_data = !use_old; break;
112 	case GIT_DELTA_DELETED:
113 		has_data = use_old; break;
114 	case GIT_DELTA_UNTRACKED:
115 		has_data = !use_old &&
116 			(diff->opts.flags & GIT_DIFF_SHOW_UNTRACKED_CONTENT) != 0;
117 		break;
118 	case GIT_DELTA_UNREADABLE:
119 	case GIT_DELTA_MODIFIED:
120 	case GIT_DELTA_COPIED:
121 	case GIT_DELTA_RENAMED:
122 		break;
123 	default:
124 		has_data = false;
125 		break;
126 	}
127 
128 	if (!has_data)
129 		fc->flags |= GIT_DIFF_FLAG__NO_DATA;
130 
131 	return diff_file_content_init_common(fc, &diff->opts);
132 }
133 
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)134 int git_diff_file_content__init_from_src(
135 	git_diff_file_content *fc,
136 	git_repository *repo,
137 	const git_diff_options *opts,
138 	const git_diff_file_content_src *src,
139 	git_diff_file *as_file)
140 {
141 	memset(fc, 0, sizeof(*fc));
142 	fc->repo = repo;
143 	fc->file = as_file;
144 
145 	if (!src->blob && !src->buf) {
146 		fc->flags |= GIT_DIFF_FLAG__NO_DATA;
147 	} else {
148 		fc->flags |= GIT_DIFF_FLAG__LOADED;
149 		fc->file->flags |= GIT_DIFF_FLAG_VALID_ID;
150 		fc->file->mode = GIT_FILEMODE_BLOB;
151 
152 		if (src->blob) {
153 			git_blob_dup((git_blob **)&fc->blob, (git_blob *) src->blob);
154 			fc->file->size = git_blob_rawsize(src->blob);
155 			git_oid_cpy(&fc->file->id, git_blob_id(src->blob));
156 			fc->file->id_abbrev = GIT_OID_HEXSZ;
157 
158 			fc->map.len  = (size_t)fc->file->size;
159 			fc->map.data = (char *)git_blob_rawcontent(src->blob);
160 
161 			fc->flags |= GIT_DIFF_FLAG__FREE_BLOB;
162 		} else {
163 			int error;
164 			if ((error = git_odb_hash(&fc->file->id, src->buf, src->buflen, GIT_OBJECT_BLOB)) < 0)
165 				return error;
166 			fc->file->size = src->buflen;
167 			fc->file->id_abbrev = GIT_OID_HEXSZ;
168 
169 			fc->map.len  = src->buflen;
170 			fc->map.data = (char *)src->buf;
171 		}
172 	}
173 
174 	return diff_file_content_init_common(fc, opts);
175 }
176 
diff_file_content_commit_to_str(git_diff_file_content * fc,bool check_status)177 static int diff_file_content_commit_to_str(
178 	git_diff_file_content *fc, bool check_status)
179 {
180 	char oid[GIT_OID_HEXSZ+1];
181 	git_buf content = GIT_BUF_INIT;
182 	const char *status = "";
183 
184 	if (check_status) {
185 		int error = 0;
186 		git_submodule *sm = NULL;
187 		unsigned int sm_status = 0;
188 		const git_oid *sm_head;
189 
190 		if ((error = git_submodule_lookup(&sm, fc->repo, fc->file->path)) < 0) {
191 			/* GIT_EEXISTS means a "submodule" that has not been git added */
192 			if (error == GIT_EEXISTS) {
193 				git_error_clear();
194 				error = 0;
195 			}
196 			return error;
197 		}
198 
199 		if ((error = git_submodule_status(&sm_status, fc->repo, fc->file->path, GIT_SUBMODULE_IGNORE_UNSPECIFIED)) < 0) {
200 			git_submodule_free(sm);
201 			return error;
202 		}
203 
204 		/* update OID if we didn't have it previously */
205 		if ((fc->file->flags & GIT_DIFF_FLAG_VALID_ID) == 0 &&
206 			((sm_head = git_submodule_wd_id(sm)) != NULL ||
207 			 (sm_head = git_submodule_head_id(sm)) != NULL))
208 		{
209 			git_oid_cpy(&fc->file->id, sm_head);
210 			fc->file->flags |= GIT_DIFF_FLAG_VALID_ID;
211 		}
212 
213 		if (GIT_SUBMODULE_STATUS_IS_WD_DIRTY(sm_status))
214 			status = "-dirty";
215 
216 		git_submodule_free(sm);
217 	}
218 
219 	git_oid_tostr(oid, sizeof(oid), &fc->file->id);
220 	if (git_buf_printf(&content, "Subproject commit %s%s\n", oid, status) < 0)
221 		return -1;
222 
223 	fc->map.len  = git_buf_len(&content);
224 	fc->map.data = git_buf_detach(&content);
225 	fc->flags |= GIT_DIFF_FLAG__FREE_DATA;
226 
227 	return 0;
228 }
229 
diff_file_content_load_blob(git_diff_file_content * fc,git_diff_options * opts)230 static int diff_file_content_load_blob(
231 	git_diff_file_content *fc,
232 	git_diff_options *opts)
233 {
234 	int error = 0;
235 	git_odb_object *odb_obj = NULL;
236 
237 	if (git_oid_is_zero(&fc->file->id))
238 		return 0;
239 
240 	if (fc->file->mode == GIT_FILEMODE_COMMIT)
241 		return diff_file_content_commit_to_str(fc, false);
242 
243 	/* if we don't know size, try to peek at object header first */
244 	if (!fc->file->size) {
245 		if ((error = git_diff_file__resolve_zero_size(
246 				fc->file, &odb_obj, fc->repo)) < 0)
247 			return error;
248 	}
249 
250 	if ((opts->flags & GIT_DIFF_SHOW_BINARY) == 0 &&
251 		diff_file_content_binary_by_size(fc))
252 		return 0;
253 
254 	if (odb_obj != NULL) {
255 		error = git_object__from_odb_object(
256 			(git_object **)&fc->blob, fc->repo, odb_obj, GIT_OBJECT_BLOB);
257 		git_odb_object_free(odb_obj);
258 	} else {
259 		error = git_blob_lookup(
260 			(git_blob **)&fc->blob, fc->repo, &fc->file->id);
261 	}
262 
263 	if (!error) {
264 		fc->flags |= GIT_DIFF_FLAG__FREE_BLOB;
265 		fc->map.data = (void *)git_blob_rawcontent(fc->blob);
266 		fc->map.len  = (size_t)git_blob_rawsize(fc->blob);
267 	}
268 
269 	return error;
270 }
271 
diff_file_content_load_workdir_symlink_fake(git_diff_file_content * fc,git_buf * path)272 static int diff_file_content_load_workdir_symlink_fake(
273 	git_diff_file_content *fc, git_buf *path)
274 {
275 	git_buf target = GIT_BUF_INIT;
276 	int error;
277 
278 	if ((error = git_futils_readbuffer(&target, path->ptr)) < 0)
279 		return error;
280 
281 	fc->map.len = git_buf_len(&target);
282 	fc->map.data = git_buf_detach(&target);
283 	fc->flags |= GIT_DIFF_FLAG__FREE_DATA;
284 
285 	git_buf_dispose(&target);
286 	return error;
287 }
288 
diff_file_content_load_workdir_symlink(git_diff_file_content * fc,git_buf * path)289 static int diff_file_content_load_workdir_symlink(
290 	git_diff_file_content *fc, git_buf *path)
291 {
292 	ssize_t alloc_len, read_len;
293 	int symlink_supported, error;
294 
295 	if ((error = git_repository__configmap_lookup(
296 		&symlink_supported, fc->repo, GIT_CONFIGMAP_SYMLINKS)) < 0)
297 		return -1;
298 
299 	if (!symlink_supported)
300 		return diff_file_content_load_workdir_symlink_fake(fc, path);
301 
302 	/* link path on disk could be UTF-16, so prepare a buffer that is
303 	 * big enough to handle some UTF-8 data expansion
304 	 */
305 	alloc_len = (ssize_t)(fc->file->size * 2) + 1;
306 
307 	fc->map.data = git__calloc(alloc_len, sizeof(char));
308 	GIT_ERROR_CHECK_ALLOC(fc->map.data);
309 
310 	fc->flags |= GIT_DIFF_FLAG__FREE_DATA;
311 
312 	read_len = p_readlink(git_buf_cstr(path), fc->map.data, alloc_len);
313 	if (read_len < 0) {
314 		git_error_set(GIT_ERROR_OS, "failed to read symlink '%s'", fc->file->path);
315 		return -1;
316 	}
317 
318 	fc->map.len = read_len;
319 	return 0;
320 }
321 
diff_file_content_load_workdir_file(git_diff_file_content * fc,git_buf * path,git_diff_options * diff_opts)322 static int diff_file_content_load_workdir_file(
323 	git_diff_file_content *fc,
324 	git_buf *path,
325 	git_diff_options *diff_opts)
326 {
327 	int error = 0;
328 	git_filter_list *fl = NULL;
329 	git_file fd = git_futils_open_ro(git_buf_cstr(path));
330 	git_buf raw = GIT_BUF_INIT;
331 
332 	if (fd < 0)
333 		return fd;
334 
335 	if (!fc->file->size)
336 	    error = git_futils_filesize(&fc->file->size, fd);
337 
338 	if (error < 0 || !fc->file->size)
339 		goto cleanup;
340 
341 	if ((diff_opts->flags & GIT_DIFF_SHOW_BINARY) == 0 &&
342 		diff_file_content_binary_by_size(fc))
343 		goto cleanup;
344 
345 	if ((error = git_filter_list_load(
346 			&fl, fc->repo, NULL, fc->file->path,
347 			GIT_FILTER_TO_ODB, GIT_FILTER_ALLOW_UNSAFE)) < 0)
348 		goto cleanup;
349 
350 	/* if there are no filters, try to mmap the file */
351 	if (fl == NULL) {
352 		if (!(error = git_futils_mmap_ro(
353 				&fc->map, fd, 0, (size_t)fc->file->size))) {
354 			fc->flags |= GIT_DIFF_FLAG__UNMAP_DATA;
355 			goto cleanup;
356 		}
357 
358 		/* if mmap failed, fall through to try readbuffer below */
359 		git_error_clear();
360 	}
361 
362 	if (!(error = git_futils_readbuffer_fd(&raw, fd, (size_t)fc->file->size))) {
363 		git_buf out = GIT_BUF_INIT;
364 
365 		error = git_filter_list__convert_buf(&out, fl, &raw);
366 
367 		if (!error) {
368 			fc->map.len  = out.size;
369 			fc->map.data = out.ptr;
370 			fc->flags |= GIT_DIFF_FLAG__FREE_DATA;
371 		}
372 	}
373 
374 cleanup:
375 	git_filter_list_free(fl);
376 	p_close(fd);
377 
378 	return error;
379 }
380 
diff_file_content_load_workdir(git_diff_file_content * fc,git_diff_options * diff_opts)381 static int diff_file_content_load_workdir(
382 	git_diff_file_content *fc,
383 	git_diff_options *diff_opts)
384 {
385 	int error = 0;
386 	git_buf path = GIT_BUF_INIT;
387 
388 	if (fc->file->mode == GIT_FILEMODE_COMMIT)
389 		return diff_file_content_commit_to_str(fc, true);
390 
391 	if (fc->file->mode == GIT_FILEMODE_TREE)
392 		return 0;
393 
394 	if (git_repository_workdir_path(&path, fc->repo, fc->file->path) < 0)
395 		return -1;
396 
397 	if (S_ISLNK(fc->file->mode))
398 		error = diff_file_content_load_workdir_symlink(fc, &path);
399 	else
400 		error = diff_file_content_load_workdir_file(fc, &path, diff_opts);
401 
402 	/* once data is loaded, update OID if we didn't have it previously */
403 	if (!error && (fc->file->flags & GIT_DIFF_FLAG_VALID_ID) == 0) {
404 		error = git_odb_hash(
405 			&fc->file->id, fc->map.data, fc->map.len, GIT_OBJECT_BLOB);
406 		fc->file->flags |= GIT_DIFF_FLAG_VALID_ID;
407 	}
408 
409 	git_buf_dispose(&path);
410 	return error;
411 }
412 
git_diff_file_content__load(git_diff_file_content * fc,git_diff_options * diff_opts)413 int git_diff_file_content__load(
414 	git_diff_file_content *fc,
415 	git_diff_options *diff_opts)
416 {
417 	int error = 0;
418 
419 	if ((fc->flags & GIT_DIFF_FLAG__LOADED) != 0)
420 		return 0;
421 
422 	if ((fc->file->flags & GIT_DIFF_FLAG_BINARY) != 0 &&
423 		(diff_opts->flags & GIT_DIFF_SHOW_BINARY) == 0)
424 		return 0;
425 
426 	if (fc->src == GIT_ITERATOR_WORKDIR)
427 		error = diff_file_content_load_workdir(fc, diff_opts);
428 	else
429 		error = diff_file_content_load_blob(fc, diff_opts);
430 	if (error)
431 		return error;
432 
433 	fc->flags |= GIT_DIFF_FLAG__LOADED;
434 
435 	diff_file_content_binary_by_content(fc);
436 
437 	return 0;
438 }
439 
git_diff_file_content__unload(git_diff_file_content * fc)440 void git_diff_file_content__unload(git_diff_file_content *fc)
441 {
442 	if ((fc->flags & GIT_DIFF_FLAG__LOADED) == 0)
443 		return;
444 
445 	if (fc->flags & GIT_DIFF_FLAG__FREE_DATA) {
446 		git__free(fc->map.data);
447 		fc->map.data = "";
448 		fc->map.len  = 0;
449 		fc->flags &= ~GIT_DIFF_FLAG__FREE_DATA;
450 	}
451 	else if (fc->flags & GIT_DIFF_FLAG__UNMAP_DATA) {
452 		git_futils_mmap_free(&fc->map);
453 		fc->map.data = "";
454 		fc->map.len  = 0;
455 		fc->flags &= ~GIT_DIFF_FLAG__UNMAP_DATA;
456 	}
457 
458 	if (fc->flags & GIT_DIFF_FLAG__FREE_BLOB) {
459 		git_blob_free((git_blob *)fc->blob);
460 		fc->blob = NULL;
461 		fc->flags &= ~GIT_DIFF_FLAG__FREE_BLOB;
462 	}
463 
464 	fc->flags &= ~GIT_DIFF_FLAG__LOADED;
465 }
466 
git_diff_file_content__clear(git_diff_file_content * fc)467 void git_diff_file_content__clear(git_diff_file_content *fc)
468 {
469 	git_diff_file_content__unload(fc);
470 
471 	/* for now, nothing else to do */
472 }
473