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 #ifndef MAX
19 #define MAX(A,B) ((A)>(B)?(A):(B))
20 #endif
21 #ifndef MIN
22 #define MIN(A,B) ((A)<(B)?(A):(B))
23 #endif
24 
25 static inline bool
26 diff_range_empty(const struct diff_range *r)
27 {
28 	return r->start == r->end;
29 }
30 
31 static inline bool
32 diff_ranges_touch(const struct diff_range *a, const struct diff_range *b)
33 {
34 	return (a->end >= b->start) && (a->start <= b->end);
35 }
36 
37 static inline void
38 diff_ranges_merge(struct diff_range *a, const struct diff_range *b)
39 {
40 	*a = (struct diff_range){
41 		.start = MIN(a->start, b->start),
42 		.end = MAX(a->end, b->end),
43 	};
44 }
45 
46 static inline int
47 diff_range_len(const struct diff_range *r)
48 {
49 	if (!r)
50 		return 0;
51 	return r->end - r->start;
52 }
53 
54 /* List of all possible return codes of a diff invocation. */
55 #define DIFF_RC_USE_DIFF_ALGO_FALLBACK	-1
56 #define DIFF_RC_OK			0
57 /* Any positive return values are errno values from sys/errno.h */
58 
59 struct diff_data;
60 
61 struct diff_atom {
62 	struct diff_data *root; /* back pointer to root diff data */
63 
64 	off_t pos;		/* if not memory-mapped */
65 	const uint8_t *at;	/* if memory-mapped */
66 	off_t len;
67 
68 	/* This hash is just a very cheap speed up for finding *mismatching*
69 	 * atoms. When hashes match, we still need to compare entire atoms to
70 	 * find out whether they are indeed identical or not. */
71 	unsigned int hash;
72 };
73 
74 int
75 diff_atom_cmp(int *cmp,
76 	      const struct diff_atom *left,
77 	      const struct diff_atom *right);
78 
79 /* Indicate whether two given diff atoms match. */
80 int
81 diff_atom_same(bool *same,
82 	       const struct diff_atom *left,
83 	       const struct diff_atom *right);
84 
85 /* The atom's index in the entire file. For atoms divided by lines of text, this
86  * yields the line number (starting with 0). Also works for diff_data that
87  * reference only a subsection of a file, always reflecting the global position
88  * in the file (and not the relative position within the subsection). */
89 #define diff_atom_root_idx(DIFF_DATA, ATOM) \
90 	((ATOM) && ((ATOM) >= (DIFF_DATA)->root->atoms.head) \
91 	 ? (unsigned int)((ATOM) - ((DIFF_DATA)->root->atoms.head)) \
92 	 : (DIFF_DATA)->root->atoms.len)
93 
94 /* The atom's index within DIFF_DATA. For atoms divided by lines of text, this
95  * yields the line number (starting with 0). */
96 #define diff_atom_idx(DIFF_DATA, ATOM) \
97 	((ATOM) && ((ATOM) >= (DIFF_DATA)->atoms.head) \
98 	 ? (unsigned int)((ATOM) - ((DIFF_DATA)->atoms.head)) \
99 	 : (DIFF_DATA)->atoms.len)
100 
101 #define foreach_diff_atom(ATOM, FIRST_ATOM, COUNT) \
102 	for ((ATOM) = (FIRST_ATOM); \
103 	     (ATOM) \
104 	     && ((ATOM) >= (FIRST_ATOM)) \
105 	     && ((ATOM) - (FIRST_ATOM) < (COUNT)); \
106 	     (ATOM)++)
107 
108 #define diff_data_foreach_atom(ATOM, DIFF_DATA) \
109 	foreach_diff_atom(ATOM, (DIFF_DATA)->atoms.head, (DIFF_DATA)->atoms.len)
110 
111 #define diff_data_foreach_atom_from(FROM, ATOM, DIFF_DATA) \
112 	for ((ATOM) = (FROM); \
113 	     (ATOM) \
114 	     && ((ATOM) >= (DIFF_DATA)->atoms.head) \
115 	     && ((ATOM) - (DIFF_DATA)->atoms.head < (DIFF_DATA)->atoms.len); \
116 	     (ATOM)++)
117 
118 /* A diff chunk represents a set of atoms on the left and/or a set of atoms on
119  * the right.
120  *
121  * If solved == false:
122  * The diff algorithm has divided the source file, and this is a chunk that the
123  * inner_algo should run on next.
124  * The lines on the left should be diffed against the lines on the right.
125  * (If there are no left lines or no right lines, it implies solved == true,
126  * because there is nothing to diff.)
127  *
128  * If solved == true:
129  * If there are only left atoms, it is a chunk removing atoms from the left ("a
130  * minus chunk").
131  * If there are only right atoms, it is a chunk adding atoms from the right ("a
132  * plus chunk").
133  * If there are both left and right lines, it is a chunk of equal content on
134  * both sides, and left_count == right_count:
135  *
136  * - foo  }
137  * - bar  }-- diff_chunk{ left_start = &left.atoms.head[0], left_count = 3,
138  * - baz  }               right_start = NULL, right_count = 0 }
139  *   moo    }
140  *   goo    }-- diff_chunk{ left_start = &left.atoms.head[3], left_count = 3,
141  *   zoo    }              right_start = &right.atoms.head[0], right_count = 3 }
142  *  +loo      }
143  *  +roo      }-- diff_chunk{ left_start = NULL, left_count = 0,
144  *  +too      }            right_start = &right.atoms.head[3], right_count = 3 }
145  *
146  */
147 struct diff_chunk {
148 	bool solved;
149 	struct diff_atom *left_start;
150 	unsigned int left_count;
151 	struct diff_atom *right_start;
152 	unsigned int right_count;
153 };
154 
155 #define DIFF_RESULT_ALLOC_BLOCKSIZE 128
156 
157 enum diff_chunk_type {
158 	CHUNK_EMPTY,
159 	CHUNK_PLUS,
160 	CHUNK_MINUS,
161 	CHUNK_SAME,
162 	CHUNK_ERROR,
163 };
164 
165 static inline enum diff_chunk_type
166 diff_chunk_type(const struct diff_chunk *chunk)
167 {
168 	if (!chunk->left_count && !chunk->right_count)
169 		return CHUNK_EMPTY;
170 	if (!chunk->solved)
171 		return CHUNK_ERROR;
172 	if (!chunk->right_count)
173 		return CHUNK_MINUS;
174 	if (!chunk->left_count)
175 		return CHUNK_PLUS;
176 	if (chunk->left_count != chunk->right_count)
177 		return CHUNK_ERROR;
178 	return CHUNK_SAME;
179 }
180 
181 struct diff_state {
182 	/* The final result passed to the original diff caller. */
183 	struct diff_result *result;
184 
185 	/* The root diff_data is in result->left,right, these are (possibly)
186 	 * subsections of the root data. */
187 	struct diff_data left;
188 	struct diff_data right;
189 
190 	unsigned int recursion_depth_left;
191 
192 	/* Remaining chunks from one diff algorithm pass, if any solved == false
193 	 * chunks came up. */
194 	diff_chunk_arraylist_t temp_result;
195 };
196 
197 struct diff_chunk *diff_state_add_chunk(struct diff_state *state, bool solved,
198 					struct diff_atom *left_start,
199 					unsigned int left_count,
200 					struct diff_atom *right_start,
201 					unsigned int right_count);
202 
203 struct diff_output_info;
204 
205 int diff_output_lines(struct diff_output_info *output_info, FILE *dest,
206 		       const char *prefix, struct diff_atom *start_atom,
207 		       unsigned int count);
208 
209 int diff_output_trailing_newline_msg(struct diff_output_info *outinfo,
210 				     FILE *dest,
211 				     const struct diff_chunk *c);
212 
213 struct diff_output_info *diff_output_info_alloc(void);
214 
215 void
216 diff_data_init_subsection(struct diff_data *d, struct diff_data *parent,
217 			  struct diff_atom *from_atom, unsigned int atoms_count);