1 /* Generic infrastructure to implement various diff algorithms. */
2 /*
3  * Copyright (c) 2020 Neels Hofmeyr <neels@hofmeyr.de>
4  *
5  * Permission to use, copy, modify, and distribute this software for any
6  * purpose with or without fee is hereby granted, provided that the above
7  * copyright notice and this permission notice appear in all copies.
8  *
9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16  */
17 
18 struct diff_range {
19 	int start;
20 	int end;
21 };
22 
23 /* List of all possible return codes of a diff invocation. */
24 #define DIFF_RC_USE_DIFF_ALGO_FALLBACK	-1
25 #define DIFF_RC_OK			0
26 /* Any positive return values are errno values from sys/errno.h */
27 
28 struct diff_atom {
29 	struct diff_data *root; /* back pointer to root diff data */
30 
31 	off_t pos;		/* if not memory-mapped */
32 	const uint8_t *at;	/* if memory-mapped */
33 	off_t len;
34 
35 	/* This hash is just a very cheap speed up for finding *mismatching*
36 	 * atoms. When hashes match, we still need to compare entire atoms to
37 	 * find out whether they are indeed identical or not.
38 	 * Calculated over all atom bytes with diff_atom_hash_update(). */
39 	unsigned int hash;
40 };
41 
42 /* Mix another atom_byte into the provided hash value and return the result.
43  * The hash value passed in for the first byte of the atom must be zero. */
44 unsigned int
45 diff_atom_hash_update(unsigned int hash, unsigned char atom_byte);
46 
47 /* Compare two atoms for equality. Return 0 on success, or errno on failure.
48  * Set cmp to -1, 0, or 1, just like strcmp(). */
49 int
50 diff_atom_cmp(int *cmp,
51 	      const struct diff_atom *left,
52 	      const struct diff_atom *right);
53 
54 
55 /* The atom's index in the entire file. For atoms divided by lines of text, this
56  * yields the line number (starting with 0). Also works for diff_data that
57  * reference only a subsection of a file, always reflecting the global position
58  * in the file (and not the relative position within the subsection). */
59 #define diff_atom_root_idx(DIFF_DATA, ATOM) \
60 	((ATOM) && ((ATOM) >= (DIFF_DATA)->root->atoms.head) \
61 	 ? (unsigned int)((ATOM) - ((DIFF_DATA)->root->atoms.head)) \
62 	 : (DIFF_DATA)->root->atoms.len)
63 
64 /* The atom's index within DIFF_DATA. For atoms divided by lines of text, this
65  * yields the line number (starting with 0). */
66 #define diff_atom_idx(DIFF_DATA, ATOM) \
67 	((ATOM) && ((ATOM) >= (DIFF_DATA)->atoms.head) \
68 	 ? (unsigned int)((ATOM) - ((DIFF_DATA)->atoms.head)) \
69 	 : (DIFF_DATA)->atoms.len)
70 
71 #define foreach_diff_atom(ATOM, FIRST_ATOM, COUNT) \
72 	for ((ATOM) = (FIRST_ATOM); \
73 	     (ATOM) \
74 	     && ((ATOM) >= (FIRST_ATOM)) \
75 	     && ((ATOM) - (FIRST_ATOM) < (COUNT)); \
76 	     (ATOM)++)
77 
78 #define diff_data_foreach_atom(ATOM, DIFF_DATA) \
79 	foreach_diff_atom(ATOM, (DIFF_DATA)->atoms.head, (DIFF_DATA)->atoms.len)
80 
81 #define diff_data_foreach_atom_from(FROM, ATOM, DIFF_DATA) \
82 	for ((ATOM) = (FROM); \
83 	     (ATOM) \
84 	     && ((ATOM) >= (DIFF_DATA)->atoms.head) \
85 	     && ((ATOM) - (DIFF_DATA)->atoms.head < (DIFF_DATA)->atoms.len); \
86 	     (ATOM)++)
87 
88 #define diff_data_foreach_atom_backwards_from(FROM, ATOM, DIFF_DATA) \
89 	for ((ATOM) = (FROM); \
90 	     (ATOM) \
91 	     && ((ATOM) >= (DIFF_DATA)->atoms.head) \
92 	     && ((ATOM) - (DIFF_DATA)->atoms.head >= 0); \
93 	     (ATOM)--)
94 
95 /* For each file, there is a "root" struct diff_data referencing the entire
96  * file, which the atoms are parsed from. In recursion of diff algorithm, there
97  * may be "child" struct diff_data only referencing a subsection of the file,
98  * re-using the atoms parsing. For "root" structs, atoms_allocated will be
99  * nonzero, indicating that the array of atoms is owned by that struct. For
100  * "child" structs, atoms_allocated == 0, to indicate that the struct is
101  * referencing a subset of atoms. */
102 struct diff_data {
103 	FILE *f;		/* if root diff_data and not memory-mapped */
104 	off_t pos;		/* if not memory-mapped */
105 	const uint8_t *data;	/* if memory-mapped */
106 	off_t len;
107 
108 	ARRAYLIST(struct diff_atom) atoms;
109 	struct diff_data *root;
110 	struct diff_data *current;
111 	void *algo_data;
112 
113 	int diff_flags;
114 
115 	int err;
116 };
117 
118 #define DIFF_FLAG_IGNORE_WHITESPACE	0x00000001
119 #define DIFF_FLAG_SHOW_PROTOTYPES	0x00000002
120 
121 void diff_data_free(struct diff_data *diff_data);
122 
123 struct diff_chunk;
124 typedef ARRAYLIST(struct diff_chunk) diff_chunk_arraylist_t;
125 
126 struct diff_result {
127 	int rc;
128 
129 	/*
130 	 * Pointers to diff data passed in via diff_main.
131 	 * Do not free these diff_data before freeing the diff_result struct.
132 	 */
133 	struct diff_data *left;
134 	struct diff_data *right;
135 
136 	diff_chunk_arraylist_t chunks;
137 };
138 
139 struct diff_state;
140 
141 /* Signature of a utility function to divide a file into diff atoms.
142  * An example is diff_atomize_text_by_line() in diff_atomize_text.c.
143  *
144  * func_data: context pointer (free to be used by implementation).
145  * d: struct diff_data with d->data and d->len already set up, and
146  * d->atoms to be created.
147  */
148 typedef int (*diff_atomize_func_t)(void *func_data, struct diff_data *d);
149 
150 extern int diff_atomize_text_by_line(void *func_data, struct diff_data *d);
151 
152 struct diff_algo_config;
153 typedef int (*diff_algo_impl_t)(
154 	const struct diff_algo_config *algo_config, struct diff_state *state);
155 
156 /* Form a result with all left-side removed and all right-side added, i.e. no
157  * actual diff algorithm involved. */
158 int diff_algo_none(const struct diff_algo_config *algo_config,
159 	struct diff_state *state);
160 
161 /* Myers Diff tracing from the start all the way through to the end, requiring
162  * quadratic amounts of memory. This can fail if the required space surpasses
163  * algo_config->permitted_state_size. */
164 extern int diff_algo_myers(const struct diff_algo_config *algo_config,
165 	struct diff_state *state);
166 
167 /* Myers "Divide et Impera": tracing forwards from the start and backwards from
168  * the end to find a midpoint that divides the problem into smaller chunks.
169  * Requires only linear amounts of memory. */
170 extern int diff_algo_myers_divide(
171 	const struct diff_algo_config *algo_config, struct diff_state *state);
172 
173 /* Patience Diff algorithm, which divides a larger diff into smaller chunks. For
174  * very specific scenarios, it may lead to a complete diff result by itself, but
175  * needs a fallback algo to solve chunks that don't have common-unique atoms. */
176 extern int diff_algo_patience(
177 	const struct diff_algo_config *algo_config, struct diff_state *state);
178 
179 /* Diff algorithms to use, possibly nested. For example:
180  *
181  * struct diff_algo_config myers, patience, myers_divide;
182  *
183  * myers = (struct diff_algo_config){
184  *         .impl = diff_algo_myers,
185  *         .permitted_state_size = 32 * 1024 * 1024,
186  *         // When too large, do diff_algo_patience:
187  *         .fallback_algo = &patience,
188  * };
189  *
190  * const struct diff_algo_config patience = (struct diff_algo_config){
191  * 	.impl = diff_algo_patience,
192  * 	// After subdivision, do Patience again:
193  * 	.inner_algo = &patience,
194  * 	// If subdivision failed, do Myers Divide et Impera:
195  * 	.fallback_algo = &myers_then_myers_divide,
196  * };
197  * 
198  * const struct diff_algo_config myers_divide = (struct diff_algo_config){
199  * 	.impl = diff_algo_myers_divide,
200  * 	// When division succeeded, start from the top:
201  * 	.inner_algo = &myers_then_myers_divide,
202  * 	// (fallback_algo = NULL implies diff_algo_none).
203  * };
204  * struct diff_config config = {
205  *         .algo = &myers,
206  *         ...
207  * };
208  * diff_main(&config, ...);
209  */
210 struct diff_algo_config {
211 	diff_algo_impl_t impl;
212 
213 	/* Fail this algo if it would use more than this amount of memory, and
214 	 * instead use fallback_algo (diff_algo_myers). permitted_state_size ==
215 	 * 0 means no limitation. */
216 	size_t permitted_state_size;
217 
218 	/* For algorithms that divide into smaller chunks, use this algorithm to
219 	 * solve the divided chunks. */
220 	const struct diff_algo_config *inner_algo;
221 
222 	/* If the algorithm fails (e.g. diff_algo_myers_if_small needs too large
223 	 * state, or diff_algo_patience can't find any common-unique atoms),
224 	 * then use this algorithm instead. */
225 	const struct diff_algo_config *fallback_algo;
226 };
227 
228 struct diff_config {
229 	diff_atomize_func_t atomize_func;
230 	void *atomize_func_data;
231 
232 	const struct diff_algo_config *algo;
233 
234 	/* How deep to step into subdivisions of a source file, a paranoia /
235 	 * safety measure to guard against infinite loops through diff
236 	 * algorithms. When the maximum recursion is reached, employ
237 	 * diff_algo_none (i.e. remove all left atoms and add all right atoms).
238 	 */
239 	unsigned int max_recursion_depth;
240 };
241 
242 int diff_atomize_file(struct diff_data *d, const struct diff_config *config,
243 		      FILE *f, const uint8_t *data, off_t len, int diff_flags);
244 struct diff_result *diff_main(const struct diff_config *config,
245 			      struct diff_data *left,
246 			      struct diff_data *right);
247 void diff_result_free(struct diff_result *result);