1 3b0f3d61 2020-01-22 neels /* Produce a unidiff output from a diff_result. */
3 3b0f3d61 2020-01-22 neels * Copyright (c) 2020 Neels Hofmeyr <neels@hofmeyr.de>
5 3b0f3d61 2020-01-22 neels * Permission to use, copy, modify, and distribute this software for any
6 3b0f3d61 2020-01-22 neels * purpose with or without fee is hereby granted, provided that the above
7 3b0f3d61 2020-01-22 neels * copyright notice and this permission notice appear in all copies.
9 3b0f3d61 2020-01-22 neels * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10 3b0f3d61 2020-01-22 neels * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11 3b0f3d61 2020-01-22 neels * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12 3b0f3d61 2020-01-22 neels * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13 3b0f3d61 2020-01-22 neels * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14 3b0f3d61 2020-01-22 neels * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15 3b0f3d61 2020-01-22 neels * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
18 e10a628a 2020-09-16 stsp #include <errno.h>
19 e10a628a 2020-09-16 stsp #include <inttypes.h>
20 e10a628a 2020-09-16 stsp #include <stdbool.h>
21 e10a628a 2020-09-16 stsp #include <stdio.h>
22 e10a628a 2020-09-16 stsp #include <stdlib.h>
24 1dfba055 2020-10-07 stsp #include <arraylist.h>
25 1dfba055 2020-10-07 stsp #include <diff_main.h>
26 1dfba055 2020-10-07 stsp #include <diff_output.h>
28 85ab4559 2020-09-22 stsp #include "diff_internal.h"
29 2a1b94d0 2020-09-26 stsp #include "diff_debug.h"
31 61a7b578 2020-05-06 neels static bool
32 f374e913 2020-09-22 stsp chunk_context_empty(const struct diff_chunk_context *cc)
34 d362ea2e 2020-07-25 stsp return diff_range_empty(&cc->chunk);
38 fe8af0d6 2020-10-06 stsp diff_chunk_get_left_start(const struct diff_chunk *c,
39 fe8af0d6 2020-10-06 stsp const struct diff_result *r, int context_lines)
41 fe8af0d6 2020-10-06 stsp int left_start = diff_atom_root_idx(&r->left, c->left_start);
42 fe8af0d6 2020-10-06 stsp return MAX(0, left_start - context_lines);
46 fe8af0d6 2020-10-06 stsp diff_chunk_get_left_end(const struct diff_chunk *c,
47 fe8af0d6 2020-10-06 stsp const struct diff_result *r, int context_lines)
49 fe8af0d6 2020-10-06 stsp int left_start = diff_chunk_get_left_start(c, r, 0);
50 fe8af0d6 2020-10-06 stsp return MIN(r->left.atoms.len,
51 fe8af0d6 2020-10-06 stsp left_start + c->left_count + context_lines);
55 fe8af0d6 2020-10-06 stsp diff_chunk_get_right_start(const struct diff_chunk *c,
56 fe8af0d6 2020-10-06 stsp const struct diff_result *r, int context_lines)
58 fe8af0d6 2020-10-06 stsp int right_start = diff_atom_root_idx(&r->right, c->right_start);
59 fe8af0d6 2020-10-06 stsp return MAX(0, right_start - context_lines);
63 fe8af0d6 2020-10-06 stsp diff_chunk_get_right_end(const struct diff_chunk *c,
64 fe8af0d6 2020-10-06 stsp const struct diff_result *r, int context_lines)
66 fe8af0d6 2020-10-06 stsp int right_start = diff_chunk_get_right_start(c, r, 0);
67 fe8af0d6 2020-10-06 stsp return MIN(r->right.atoms.len,
68 fe8af0d6 2020-10-06 stsp right_start + c->right_count + context_lines);
72 f374e913 2020-09-22 stsp diff_chunk_context_get(struct diff_chunk_context *cc, const struct diff_result *r,
73 0d27172a 2020-05-06 neels int chunk_idx, int context_lines)
75 3b0f3d61 2020-01-22 neels const struct diff_chunk *c = &r->chunks.head[chunk_idx];
76 fe8af0d6 2020-10-06 stsp int left_start = diff_chunk_get_left_start(c, r, context_lines);
77 fe8af0d6 2020-10-06 stsp int left_end = diff_chunk_get_left_end(c, r, context_lines);
78 fe8af0d6 2020-10-06 stsp int right_start = diff_chunk_get_right_start(c, r, context_lines);
79 fe8af0d6 2020-10-06 stsp int right_end = diff_chunk_get_right_end(c, r, context_lines);
81 f374e913 2020-09-22 stsp *cc = (struct diff_chunk_context){
83 3b0f3d61 2020-01-22 neels .start = chunk_idx,
84 3b0f3d61 2020-01-22 neels .end = chunk_idx + 1,
87 0d27172a 2020-05-06 neels .start = left_start,
88 0d27172a 2020-05-06 neels .end = left_end,
91 0d27172a 2020-05-06 neels .start = right_start,
92 0d27172a 2020-05-06 neels .end = right_end,
97 61a7b578 2020-05-06 neels static bool
98 f374e913 2020-09-22 stsp chunk_contexts_touch(const struct diff_chunk_context *cc,
99 f374e913 2020-09-22 stsp const struct diff_chunk_context *other)
101 d362ea2e 2020-07-25 stsp return diff_ranges_touch(&cc->chunk, &other->chunk)
102 d362ea2e 2020-07-25 stsp || diff_ranges_touch(&cc->left, &other->left)
103 d362ea2e 2020-07-25 stsp || diff_ranges_touch(&cc->right, &other->right);
106 61a7b578 2020-05-06 neels static void
107 f374e913 2020-09-22 stsp chunk_contexts_merge(struct diff_chunk_context *cc,
108 f374e913 2020-09-22 stsp const struct diff_chunk_context *other)
110 d362ea2e 2020-07-25 stsp diff_ranges_merge(&cc->chunk, &other->chunk);
111 d362ea2e 2020-07-25 stsp diff_ranges_merge(&cc->left, &other->left);
112 d362ea2e 2020-07-25 stsp diff_ranges_merge(&cc->right, &other->right);
115 11caa5cc 2020-09-22 stsp struct diff_output_unidiff_state {
116 11caa5cc 2020-09-22 stsp bool header_printed;
119 11caa5cc 2020-09-22 stsp struct diff_output_unidiff_state *
120 11caa5cc 2020-09-22 stsp diff_output_unidiff_state_alloc(void)
122 11caa5cc 2020-09-22 stsp struct diff_output_unidiff_state *state;
124 11caa5cc 2020-09-22 stsp state = calloc(1, sizeof(struct diff_output_unidiff_state));
125 11caa5cc 2020-09-22 stsp if (state != NULL)
126 11caa5cc 2020-09-22 stsp diff_output_unidiff_state_reset(state);
127 11caa5cc 2020-09-22 stsp return state;
131 11caa5cc 2020-09-22 stsp diff_output_unidiff_state_reset(struct diff_output_unidiff_state *state)
133 11caa5cc 2020-09-22 stsp state->header_printed = false;
137 11caa5cc 2020-09-22 stsp diff_output_unidiff_state_free(struct diff_output_unidiff_state *state)
139 11caa5cc 2020-09-22 stsp free(state);
143 2c20a3ed 2020-09-22 stsp output_unidiff_chunk(struct diff_output_info *outinfo, FILE *dest,
144 2c20a3ed 2020-09-22 stsp struct diff_output_unidiff_state *state,
145 2c20a3ed 2020-09-22 stsp const struct diff_input_info *info,
146 2c20a3ed 2020-09-22 stsp const struct diff_result *result,
147 2c20a3ed 2020-09-22 stsp const struct diff_chunk_context *cc)
149 b6adedb6 2020-10-07 stsp int rc, left_start, left_len, right_start, right_len;
150 ab528e22 2020-09-22 stsp off_t outoff = 0, *offp;
152 d362ea2e 2020-07-25 stsp if (diff_range_empty(&cc->left) && diff_range_empty(&cc->right))
153 2c20a3ed 2020-09-22 stsp return DIFF_RC_OK;
155 2c20a3ed 2020-09-22 stsp if (outinfo && outinfo->line_offsets.len > 0) {
156 2c20a3ed 2020-09-22 stsp unsigned int idx = outinfo->line_offsets.len - 1;
157 2c20a3ed 2020-09-22 stsp outoff = outinfo->line_offsets.head[idx];
160 11caa5cc 2020-09-22 stsp if (!(state->header_printed)) {
161 2c20a3ed 2020-09-22 stsp rc = fprintf(dest, "--- %s\n", info->left_path ? : "a");
162 2c20a3ed 2020-09-22 stsp if (rc < 0)
163 2c20a3ed 2020-09-22 stsp return errno;
164 2c20a3ed 2020-09-22 stsp if (outinfo) {
165 2c20a3ed 2020-09-22 stsp ARRAYLIST_ADD(offp, outinfo->line_offsets);
166 dabc1008 2020-09-22 stsp if (offp == NULL)
167 dabc1008 2020-09-22 stsp return ENOMEM;
168 2c20a3ed 2020-09-22 stsp outoff += rc;
169 2c20a3ed 2020-09-22 stsp *offp = outoff;
172 2c20a3ed 2020-09-22 stsp rc = fprintf(dest, "+++ %s\n", info->right_path ? : "b");
173 2c20a3ed 2020-09-22 stsp if (rc < 0)
174 2c20a3ed 2020-09-22 stsp return errno;
175 2c20a3ed 2020-09-22 stsp if (outinfo) {
176 2c20a3ed 2020-09-22 stsp ARRAYLIST_ADD(offp, outinfo->line_offsets);
177 dabc1008 2020-09-22 stsp if (offp == NULL)
178 dabc1008 2020-09-22 stsp return ENOMEM;
179 2c20a3ed 2020-09-22 stsp outoff += rc;
180 2c20a3ed 2020-09-22 stsp *offp = outoff;
183 11caa5cc 2020-09-22 stsp state->header_printed = true;
186 b6adedb6 2020-10-07 stsp left_len = cc->left.end - cc->left.start;
187 11d9f2f7 2020-10-07 stsp if (result->left.atoms.len == 0)
188 11d9f2f7 2020-10-07 stsp left_start = 0;
189 11d9f2f7 2020-10-07 stsp else if (left_len == 0 && cc->left.start > 0)
190 b6adedb6 2020-10-07 stsp left_start = cc->left.start;
192 b6adedb6 2020-10-07 stsp left_start = cc->left.start + 1;
194 b6adedb6 2020-10-07 stsp right_len = cc->right.end - cc->right.start;
195 11d9f2f7 2020-10-07 stsp if (result->right.atoms.len == 0)
196 11d9f2f7 2020-10-07 stsp right_start = 0;
197 11d9f2f7 2020-10-07 stsp else if (right_len == 0 && cc->right.start > 0)
198 b6adedb6 2020-10-07 stsp right_start = cc->right.start;
200 b6adedb6 2020-10-07 stsp right_start = cc->right.start + 1;
202 b6adedb6 2020-10-07 stsp if (left_len == 1 && right_len != 1) {
203 b6adedb6 2020-10-07 stsp rc = fprintf(dest, "@@ -%d +%d,%d @@\n",
204 b6adedb6 2020-10-07 stsp left_start, right_start, right_len);
205 b6adedb6 2020-10-07 stsp } else if (left_len != 1 && right_len == 1) {
206 b6adedb6 2020-10-07 stsp rc = fprintf(dest, "@@ -%d,%d +%d @@\n",
207 b6adedb6 2020-10-07 stsp left_start, left_len, right_start);
209 b6adedb6 2020-10-07 stsp rc = fprintf(dest, "@@ -%d,%d +%d,%d @@\n",
210 b6adedb6 2020-10-07 stsp left_start, left_len, right_start, right_len);
212 2c20a3ed 2020-09-22 stsp if (rc < 0)
213 2c20a3ed 2020-09-22 stsp return errno;
214 2c20a3ed 2020-09-22 stsp if (outinfo) {
215 2c20a3ed 2020-09-22 stsp ARRAYLIST_ADD(offp, outinfo->line_offsets);
216 dabc1008 2020-09-22 stsp if (offp == NULL)
217 dabc1008 2020-09-22 stsp return ENOMEM;
218 2c20a3ed 2020-09-22 stsp outoff += rc;
219 2c20a3ed 2020-09-22 stsp *offp = outoff;
223 0d27172a 2020-05-06 neels /* Got the absolute line numbers where to start printing, and the index
224 0d27172a 2020-05-06 neels * of the interesting (non-context) chunk.
225 0d27172a 2020-05-06 neels * To print context lines above the interesting chunk, nipping on the
226 0d27172a 2020-05-06 neels * previous chunk index may be necessary.
227 0d27172a 2020-05-06 neels * It is guaranteed to be only context lines where left == right, so it
228 0d27172a 2020-05-06 neels * suffices to look on the left. */
229 0d27172a 2020-05-06 neels const struct diff_chunk *first_chunk;
230 0d27172a 2020-05-06 neels int chunk_start_line;
231 0d27172a 2020-05-06 neels first_chunk = &result->chunks.head[cc->chunk.start];
232 0d27172a 2020-05-06 neels chunk_start_line = diff_atom_root_idx(&result->left,
233 0d27172a 2020-05-06 neels first_chunk->left_start);
234 2c20a3ed 2020-09-22 stsp if (cc->left.start < chunk_start_line) {
235 2c20a3ed 2020-09-22 stsp rc = diff_output_lines(outinfo, dest, " ",
236 0d27172a 2020-05-06 neels &result->left.atoms.head[cc->left.start],
237 3b0f3d61 2020-01-22 neels chunk_start_line - cc->left.start);
242 3b0f3d61 2020-01-22 neels /* Now write out all the joined chunks and contexts between them */
243 3b0f3d61 2020-01-22 neels int c_idx;
244 3b0f3d61 2020-01-22 neels for (c_idx = cc->chunk.start; c_idx < cc->chunk.end; c_idx++) {
245 3b0f3d61 2020-01-22 neels const struct diff_chunk *c = &result->chunks.head[c_idx];
247 3b0f3d61 2020-01-22 neels if (c->left_count && c->right_count)
248 2c20a3ed 2020-09-22 stsp rc = diff_output_lines(outinfo, dest,
249 0d27172a 2020-05-06 neels c->solved ? " " : "?",
250 0d27172a 2020-05-06 neels c->left_start, c->left_count);
251 3b0f3d61 2020-01-22 neels else if (c->left_count && !c->right_count)
252 2c20a3ed 2020-09-22 stsp rc = diff_output_lines(outinfo, dest,
253 0d27172a 2020-05-06 neels c->solved ? "-" : "?",
254 0d27172a 2020-05-06 neels c->left_start, c->left_count);
255 3b0f3d61 2020-01-22 neels else if (c->right_count && !c->left_count)
256 2c20a3ed 2020-09-22 stsp rc = diff_output_lines(outinfo, dest,
257 0d27172a 2020-05-06 neels c->solved ? "+" : "?",
258 0d27172a 2020-05-06 neels c->right_start, c->right_count);
263 3b0f3d61 2020-01-22 neels /* Trailing context? */
264 0d27172a 2020-05-06 neels const struct diff_chunk *last_chunk;
265 0d27172a 2020-05-06 neels int chunk_end_line;
266 0d27172a 2020-05-06 neels last_chunk = &result->chunks.head[cc->chunk.end - 1];
267 0d27172a 2020-05-06 neels chunk_end_line = diff_atom_root_idx(&result->left,
268 0d27172a 2020-05-06 neels last_chunk->left_start
269 0d27172a 2020-05-06 neels + last_chunk->left_count);
270 2c20a3ed 2020-09-22 stsp if (cc->left.end > chunk_end_line) {
271 2c20a3ed 2020-09-22 stsp rc = diff_output_lines(outinfo, dest, " ",
272 0d27172a 2020-05-06 neels &result->left.atoms.head[chunk_end_line],
273 3b0f3d61 2020-01-22 neels cc->left.end - chunk_end_line);
278 2c20a3ed 2020-09-22 stsp return DIFF_RC_OK;
282 2c20a3ed 2020-09-22 stsp diff_output_unidiff_chunk(struct diff_output_info **output_info, FILE *dest,
283 2c20a3ed 2020-09-22 stsp struct diff_output_unidiff_state *state,
284 2c20a3ed 2020-09-22 stsp const struct diff_input_info *info,
285 2c20a3ed 2020-09-22 stsp const struct diff_result *result,
286 2c20a3ed 2020-09-22 stsp const struct diff_chunk_context *cc)
288 2c20a3ed 2020-09-22 stsp struct diff_output_info *outinfo = NULL;
290 2c20a3ed 2020-09-22 stsp if (output_info) {
291 2c20a3ed 2020-09-22 stsp *output_info = diff_output_info_alloc();
292 2c20a3ed 2020-09-22 stsp if (*output_info == NULL)
293 2c20a3ed 2020-09-22 stsp return ENOMEM;
294 2c20a3ed 2020-09-22 stsp outinfo = *output_info;
297 2c20a3ed 2020-09-22 stsp return output_unidiff_chunk(outinfo, dest, state, info,
298 2c20a3ed 2020-09-22 stsp result, cc);
302 2c20a3ed 2020-09-22 stsp diff_output_unidiff(struct diff_output_info **output_info,
303 2c20a3ed 2020-09-22 stsp FILE *dest, const struct diff_input_info *info,
304 0d27172a 2020-05-06 neels const struct diff_result *result,
305 0d27172a 2020-05-06 neels unsigned int context_lines)
307 11caa5cc 2020-09-22 stsp struct diff_output_unidiff_state *state;
308 11caa5cc 2020-09-22 stsp struct diff_chunk_context cc = {};
309 2c20a3ed 2020-09-22 stsp struct diff_output_info *outinfo = NULL;
312 3b0f3d61 2020-01-22 neels if (!result)
313 3e6cba3a 2020-08-13 stsp return EINVAL;
314 3b0f3d61 2020-01-22 neels if (result->rc != DIFF_RC_OK)
315 3b0f3d61 2020-01-22 neels return result->rc;
317 2c20a3ed 2020-09-22 stsp if (output_info) {
318 2c20a3ed 2020-09-22 stsp *output_info = diff_output_info_alloc();
319 2c20a3ed 2020-09-22 stsp if (*output_info == NULL)
320 2c20a3ed 2020-09-22 stsp return ENOMEM;
321 2c20a3ed 2020-09-22 stsp outinfo = *output_info;
324 11caa5cc 2020-09-22 stsp state = diff_output_unidiff_state_alloc();
325 2c20a3ed 2020-09-22 stsp if (state == NULL) {
326 2c20a3ed 2020-09-22 stsp if (output_info) {
327 2c20a3ed 2020-09-22 stsp diff_output_info_free(*output_info);
328 2c20a3ed 2020-09-22 stsp *output_info = NULL;
330 11caa5cc 2020-09-22 stsp return ENOMEM;
334 3b0f3d61 2020-01-22 neels for (i = 0; i < result->chunks.len; i++) {
335 3b0f3d61 2020-01-22 neels struct diff_chunk *c = &result->chunks.head[i];
336 8546b045 2020-09-20 neels enum diff_chunk_type t = diff_chunk_type(c);
337 f374e913 2020-09-22 stsp struct diff_chunk_context next;
339 9cc49695 2020-05-05 neels if (t != CHUNK_MINUS && t != CHUNK_PLUS)
342 9cc49695 2020-05-05 neels if (chunk_context_empty(&cc)) {
343 9cc49695 2020-05-05 neels /* These are the first lines being printed.
344 0d27172a 2020-05-06 neels * Note down the start point, any number of subsequent
345 0d27172a 2020-05-06 neels * chunks may be joined up to this unidiff chunk by
346 0d27172a 2020-05-06 neels * context lines or by being directly adjacent. */
347 f374e913 2020-09-22 stsp diff_chunk_context_get(&cc, result, i, context_lines);
348 0d27172a 2020-05-06 neels debug("new chunk to be printed:"
349 0d27172a 2020-05-06 neels " chunk %d-%d left %d-%d right %d-%d\n",
350 9cc49695 2020-05-05 neels cc.chunk.start, cc.chunk.end,
351 9cc49695 2020-05-05 neels cc.left.start, cc.left.end,
352 9cc49695 2020-05-05 neels cc.right.start, cc.right.end);
356 0d27172a 2020-05-06 neels /* There already is a previous chunk noted down for being
357 0d27172a 2020-05-06 neels * printed. Does it join up with this one? */
358 f374e913 2020-09-22 stsp diff_chunk_context_get(&next, result, i, context_lines);
359 0d27172a 2020-05-06 neels debug("new chunk to be printed:"
360 0d27172a 2020-05-06 neels " chunk %d-%d left %d-%d right %d-%d\n",
361 9cc49695 2020-05-05 neels next.chunk.start, next.chunk.end,
362 9cc49695 2020-05-05 neels next.left.start, next.left.end,
363 9cc49695 2020-05-05 neels next.right.start, next.right.end);
365 9cc49695 2020-05-05 neels if (chunk_contexts_touch(&cc, &next)) {
366 0d27172a 2020-05-06 neels /* This next context touches or overlaps the previous
367 0d27172a 2020-05-06 neels * one, join. */
368 9cc49695 2020-05-05 neels chunk_contexts_merge(&cc, &next);
369 0d27172a 2020-05-06 neels debug("new chunk to be printed touches previous chunk,"
370 0d27172a 2020-05-06 neels " now: left %d-%d right %d-%d\n",
371 9cc49695 2020-05-05 neels cc.left.start, cc.left.end,
372 9cc49695 2020-05-05 neels cc.right.start, cc.right.end);
376 0d27172a 2020-05-06 neels /* No touching, so the previous context is complete with a gap
377 0d27172a 2020-05-06 neels * between it and this next one. Print the previous one and
378 0d27172a 2020-05-06 neels * start fresh here. */
379 0d27172a 2020-05-06 neels debug("new chunk to be printed does not touch previous chunk;"
380 0d27172a 2020-05-06 neels " print left %d-%d right %d-%d\n",
381 9cc49695 2020-05-05 neels cc.left.start, cc.left.end, cc.right.start, cc.right.end);
382 2c20a3ed 2020-09-22 stsp output_unidiff_chunk(outinfo, dest, state, info, result, &cc);
383 9cc49695 2020-05-05 neels cc = next;
384 9cc49695 2020-05-05 neels debug("new unprinted chunk is left %d-%d right %d-%d\n",
385 9cc49695 2020-05-05 neels cc.left.start, cc.left.end, cc.right.start, cc.right.end);
388 3b0f3d61 2020-01-22 neels if (!chunk_context_empty(&cc))
389 2c20a3ed 2020-09-22 stsp output_unidiff_chunk(outinfo, dest, state, info, result, &cc);
390 11caa5cc 2020-09-22 stsp diff_output_unidiff_state_free(state);
391 3b0f3d61 2020-01-22 neels return DIFF_RC_OK;