source src/trailer.c
Line | Flow | Count | Block(s) | Source |
---|---|---|---|---|
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 "array.h" | ||
8 | - | #include "common.h" | ||
9 | - | #include "git2/message.h" | ||
10 | - | |||
11 | - | #include <stddef.h> | ||
12 | - | #include <string.h> | ||
13 | - | #include <ctype.h> | ||
14 | - | |||
15 | - | #define COMMENT_LINE_CHAR '#' | ||
16 | - | #define TRAILER_SEPARATORS ":" | ||
17 | - | |||
18 | - | static const char *const git_generated_prefixes[] = { | ||
19 | - | "Signed-off-by: ", | ||
20 | - | "(cherry picked from commit ", | ||
21 | - | NULL | ||
22 | - | }; | ||
23 | - | |||
24 | 43 | 2 | static int is_blank_line(const char *str) | |
25 | - | { | ||
26 | 43 | 2 | const char *s = str; | |
27 | 45 | 2,4-7 | while (*s && *s != '\n' && isspace(*s)) | |
28 | 2 | 3 | s++; | |
29 | 43 | 8 | return !*s || *s == '\n'; | |
30 | - | } | ||
31 | - | |||
32 | 57 | 2 | static const char *next_line(const char *str) | |
33 | - | { | ||
34 | 57 | 2 | const char *nl = strchr(str, '\n'); | |
35 | - | |||
36 | 57 | 2 | if (nl) { | |
37 | 56 | 3 | return nl + 1; | |
38 | - | } else { | ||
39 | - | /* return pointer to the NUL terminator: */ | ||
40 | 1 | 4 | return str + strlen(str); | |
41 | - | } | ||
42 | - | } | ||
43 | - | |||
44 | - | /* | ||
45 | - | * Return the position of the start of the last line. If len is 0, return 0. | ||
46 | - | */ | ||
47 | 25 | 2 | static bool last_line(size_t *out, const char *buf, size_t len) | |
48 | - | { | ||
49 | - | size_t i; | ||
50 | - | |||
51 | 25 | 2 | *out = 0; | |
52 | - | |||
53 | 25 | 2 | if (len == 0) | |
54 | ##### | 3 | return false; | |
55 | 25 | 4 | if (len == 1) | |
56 | ##### | 5 | return true; | |
57 | - | |||
58 | - | /* | ||
59 | - | * Skip the last character (in addition to the null terminator), | ||
60 | - | * because if the last character is a newline, it is considered as part | ||
61 | - | * of the last line anyway. | ||
62 | - | */ | ||
63 | 25 | 6 | i = len - 2; | |
64 | - | |||
65 | 222 | 6,9,10 | for (; i > 0; i--) { | |
66 | 222 | 7 | if (buf[i] == '\n') { | |
67 | 25 | 8 | *out = i + 1; | |
68 | 25 | 8 | return true; | |
69 | - | } | ||
70 | - | } | ||
71 | ##### | 11 | return true; | |
72 | - | } | ||
73 | - | |||
74 | - | /* | ||
75 | - | * If the given line is of the form | ||
76 | - | * "<token><optional whitespace><separator>..." or "<separator>...", sets out | ||
77 | - | * to the location of the separator and returns true. Otherwise, returns | ||
78 | - | * false. The optional whitespace is allowed there primarily to allow things | ||
79 | - | * like "Bug #43" where <token> is "Bug" and <separator> is "#". | ||
80 | - | * | ||
81 | - | * The separator-starts-line case (in which this function returns true and | ||
82 | - | * sets out to 0) is distinguished from the non-well-formed-line case (in | ||
83 | - | * which this function returns false) because some callers of this function | ||
84 | - | * need such a distinction. | ||
85 | - | */ | ||
86 | 13 | 2 | static bool find_separator(size_t *out, const char *line, const char *separators) | |
87 | - | { | ||
88 | 13 | 2 | int whitespace_found = 0; | |
89 | - | const char *c; | ||
90 | 50 | 2,15,16 | for (c = line; *c; c++) { | |
91 | 50 | 3 | if (strchr(separators, *c)) { | |
92 | 9 | 4 | *out = c - line; | |
93 | 9 | 4 | return true; | |
94 | - | } | ||
95 | - | |||
96 | 41 | 5-8 | if (!whitespace_found && (isalnum(*c) || *c == '-')) | |
97 | 32 | 9 | continue; | |
98 | 9 | 10-12 | if (c != line && (*c == ' ' || *c == '\t')) { | |
99 | 5 | 13 | whitespace_found = 1; | |
100 | 5 | 13 | continue; | |
101 | - | } | ||
102 | 4 | 14 | break; | |
103 | - | } | ||
104 | 4 | 17 | return false; | |
105 | - | } | ||
106 | - | |||
107 | - | /* | ||
108 | - | * Inspect the given string and determine the true "end" of the log message, in | ||
109 | - | * order to find where to put a new Signed-off-by: line. Ignored are | ||
110 | - | * trailing comment lines and blank lines. To support "git commit -s | ||
111 | - | * --amend" on an existing commit, we also ignore "Conflicts:". To | ||
112 | - | * support "git commit -v", we truncate at cut lines. | ||
113 | - | * | ||
114 | - | * Returns the number of bytes from the tail to ignore, to be fed as | ||
115 | - | * the second parameter to append_signoff(). | ||
116 | - | */ | ||
117 | 9 | 2 | static size_t ignore_non_trailer(const char *buf, size_t len) | |
118 | - | { | ||
119 | 9 | 2 | size_t boc = 0, bol = 0; | |
120 | 9 | 2 | int in_old_conflicts_block = 0; | |
121 | 9 | 2 | size_t cutoff = len; | |
122 | - | |||
123 | 49 | 2,22 | while (bol < cutoff) { | |
124 | 40 | 3 | const char *next_line = memchr(buf + bol, '\n', len - bol); | |
125 | - | |||
126 | 40 | 3 | if (!next_line) | |
127 | 1 | 4 | next_line = buf + len; | |
128 | - | else | ||
129 | 39 | 5 | next_line++; | |
130 | - | |||
131 | 40 | 6,7 | if (buf[bol] == COMMENT_LINE_CHAR || buf[bol] == '\n') { | |
132 | - | /* is this the first of the run of comments? */ | ||
133 | 12 | 8,10 | if (!boc) | |
134 | 12 | 9 | boc = bol; | |
135 | - | /* otherwise, it is just continuing */ | ||
136 | 28 | 11,12 | } else if (git__prefixcmp(buf + bol, "Conflicts:\n") == 0) { | |
137 | 1 | 13 | in_old_conflicts_block = 1; | |
138 | 1 | 13 | if (!boc) | |
139 | 1 | 14,15 | boc = bol; | |
140 | 27 | 16-18 | } else if (in_old_conflicts_block && buf[bol] == '\t') { | |
141 | - | ; /* a pathname in the conflicts block */ | ||
142 | 26 | 19 | } else if (boc) { | |
143 | - | /* the previous was not trailing comment */ | ||
144 | 10 | 20 | boc = 0; | |
145 | 10 | 20 | in_old_conflicts_block = 0; | |
146 | - | } | ||
147 | 40 | 21 | bol = next_line - buf; | |
148 | - | } | ||
149 | 9 | 23 | return boc ? len - boc : len - cutoff; | |
150 | - | } | ||
151 | - | |||
152 | - | /* | ||
153 | - | * Return the position of the start of the patch or the length of str if there | ||
154 | - | * is no patch in the message. | ||
155 | - | */ | ||
156 | 9 | 2 | static size_t find_patch_start(const char *str) | |
157 | - | { | ||
158 | - | const char *s; | ||
159 | - | |||
160 | 49 | 2,6,7 | for (s = str; *s; s = next_line(s)) { | |
161 | 41 | 3,4 | if (git__prefixcmp(s, "---") == 0) | |
162 | 1 | 5 | return s - str; | |
163 | - | } | ||
164 | - | |||
165 | 8 | 8 | return s - str; | |
166 | - | } | ||
167 | - | |||
168 | - | /* | ||
169 | - | * Return the position of the first trailer line or len if there are no | ||
170 | - | * trailers. | ||
171 | - | */ | ||
172 | 9 | 2 | static size_t find_trailer_start(const char *buf, size_t len) | |
173 | - | { | ||
174 | - | const char *s; | ||
175 | - | size_t end_of_title, l; | ||
176 | 9 | 2 | int only_spaces = 1; | |
177 | 9 | 2 | int recognized_prefix = 0, trailer_lines = 0, non_trailer_lines = 0; | |
178 | - | /* | ||
179 | - | * Number of possible continuation lines encountered. This will be | ||
180 | - | * reset to 0 if we encounter a trailer (since those lines are to be | ||
181 | - | * considered continuations of that trailer), and added to | ||
182 | - | * non_trailer_lines if we encounter a non-trailer (since those lines | ||
183 | - | * are to be considered non-trailers). | ||
184 | - | */ | ||
185 | 9 | 2 | int possible_continuation_lines = 0; | |
186 | - | |||
187 | - | /* The first paragraph is the title and cannot be trailers */ | ||
188 | 18 | 2,8,9 | for (s = buf; s < buf + len; s = next_line(s)) { | |
189 | 18 | 3 | if (s[0] == COMMENT_LINE_CHAR) | |
190 | ##### | 4 | continue; | |
191 | 18 | 5,6 | if (is_blank_line(s)) | |
192 | 9 | 7 | break; | |
193 | - | } | ||
194 | 9 | 10 | end_of_title = s - buf; | |
195 | - | |||
196 | - | /* | ||
197 | - | * Get the start of the trailers by looking starting from the end for a | ||
198 | - | * blank line before a set of non-blank lines that (i) are all | ||
199 | - | * trailers, or (ii) contains at least one Git-generated trailer and | ||
200 | - | * consists of at least 25% trailers. | ||
201 | - | */ | ||
202 | 9 | 10 | l = len; | |
203 | 25 | 10,46-48 | while (last_line(&l, buf, l) && l >= end_of_title) { | |
204 | 25 | 11 | const char *bol = buf + l; | |
205 | - | const char *const *p; | ||
206 | 25 | 11 | size_t separator_pos = 0; | |
207 | - | |||
208 | 25 | 11 | if (bol[0] == COMMENT_LINE_CHAR) { | |
209 | ##### | 12 | non_trailer_lines += possible_continuation_lines; | |
210 | ##### | 12 | possible_continuation_lines = 0; | |
211 | ##### | 12,44 | continue; | |
212 | - | } | ||
213 | 25 | 13,14 | if (is_blank_line(bol)) { | |
214 | 9 | 15 | if (only_spaces) | |
215 | ##### | 16 | continue; | |
216 | 9 | 17 | non_trailer_lines += possible_continuation_lines; | |
217 | 9 | 17,18 | if (recognized_prefix && | |
218 | 2 | 18 | trailer_lines * 3 >= non_trailer_lines) | |
219 | 9 | 19,20,45 | return next_line(bol) - buf; | |
220 | 7 | 21,22 | else if (trailer_lines && !non_trailer_lines) | |
221 | 6 | 23,24 | return next_line(bol) - buf; | |
222 | 1 | 25 | return len; | |
223 | - | } | ||
224 | 16 | 26 | only_spaces = 0; | |
225 | - | |||
226 | 42 | 26,30,31 | for (p = git_generated_prefixes; *p; p++) { | |
227 | 29 | 27,28 | if (git__prefixcmp(bol, *p) == 0) { | |
228 | 3 | 29 | trailer_lines++; | |
229 | 3 | 29 | possible_continuation_lines = 0; | |
230 | 3 | 29 | recognized_prefix = 1; | |
231 | 3 | 29 | goto continue_outer_loop; | |
232 | - | } | ||
233 | - | } | ||
234 | - | |||
235 | 13 | 32 | find_separator(&separator_pos, bol, TRAILER_SEPARATORS); | |
236 | 13 | 33-35 | if (separator_pos >= 1 && !isspace(bol[0])) { | |
237 | 9 | 36 | trailer_lines++; | |
238 | 9 | 36 | possible_continuation_lines = 0; | |
239 | 9 | 36,38 | if (recognized_prefix) | |
240 | ##### | 37 | continue; | |
241 | 4 | 39,40 | } else if (isspace(bol[0])) | |
242 | 2 | 41 | possible_continuation_lines++; | |
243 | - | else { | ||
244 | 2 | 42 | non_trailer_lines++; | |
245 | 2 | 42 | non_trailer_lines += possible_continuation_lines; | |
246 | 16 | 42,43 | possible_continuation_lines = 0; | |
247 | - | } | ||
248 | - | continue_outer_loop: | ||
249 | - | ; | ||
250 | - | } | ||
251 | - | |||
252 | ##### | 49 | return len; | |
253 | - | } | ||
254 | - | |||
255 | - | /* Return the position of the end of the trailers. */ | ||
256 | 9 | 2 | static size_t find_trailer_end(const char *buf, size_t len) | |
257 | - | { | ||
258 | 9 | 2 | return len - ignore_non_trailer(buf, len); | |
259 | - | } | ||
260 | - | |||
261 | 9 | 2 | static char *extract_trailer_block(const char *message, size_t* len) | |
262 | - | { | ||
263 | 9 | 2 | size_t patch_start = find_patch_start(message); | |
264 | 9 | 3 | size_t trailer_end = find_trailer_end(message, patch_start); | |
265 | 9 | 4 | size_t trailer_start = find_trailer_start(message, trailer_end); | |
266 | - | |||
267 | 9 | 5 | size_t trailer_len = trailer_end - trailer_start; | |
268 | - | |||
269 | 9 | 5 | char *buffer = git__malloc(trailer_len + 1); | |
270 | 9 | 6 | if (buffer == NULL) | |
271 | ##### | 7 | return NULL; | |
272 | - | |||
273 | 9 | 8 | memcpy(buffer, message + trailer_start, trailer_len); | |
274 | 9 | 8 | buffer[trailer_len] = 0; | |
275 | - | |||
276 | 9 | 8 | *len = trailer_len; | |
277 | - | |||
278 | 9 | 8 | return buffer; | |
279 | - | } | ||
280 | - | |||
281 | - | enum trailer_state { | ||
282 | - | S_START = 0, | ||
283 | - | S_KEY = 1, | ||
284 | - | S_KEY_WS = 2, | ||
285 | - | S_SEP_WS = 3, | ||
286 | - | S_VALUE = 4, | ||
287 | - | S_VALUE_NL = 5, | ||
288 | - | S_VALUE_END = 6, | ||
289 | - | S_IGNORE = 7, | ||
290 | - | }; | ||
291 | - | |||
292 | - | #define NEXT(st) { state = (st); ptr++; continue; } | ||
293 | - | #define GOTO(st) { state = (st); continue; } | ||
294 | - | |||
295 | - | typedef git_array_t(git_message_trailer) git_array_trailer_t; | ||
296 | - | |||
297 | 9 | 2 | int git_message_trailers(git_message_trailer_array *trailer_arr, const char *message) | |
298 | - | { | ||
299 | 9 | 2 | enum trailer_state state = S_START; | |
300 | 9 | 2 | int rc = 0; | |
301 | - | char *ptr; | ||
302 | 9 | 2 | char *key = NULL; | |
303 | 9 | 2 | char *value = NULL; | |
304 | 9 | 2 | git_array_trailer_t arr = GIT_ARRAY_INIT; | |
305 | - | |||
306 | - | size_t trailer_len; | ||
307 | 9 | 2 | char *trailer = extract_trailer_block(message, &trailer_len); | |
308 | 9 | 3 | if (trailer == NULL) | |
309 | ##### | 4 | return -1; | |
310 | - | |||
311 | 9 | 5 | for (ptr = trailer;;) { | |
312 | 249 | 6 | switch (state) { | |
313 | - | case S_START: { | ||
314 | 22 | 7 | if (*ptr == 0) { | |
315 | 9 | 8 | goto ret; | |
316 | - | } | ||
317 | - | |||
318 | 13 | 9 | key = ptr; | |
319 | 13 | 9 | GOTO(S_KEY); | |
320 | - | } | ||
321 | - | case S_KEY: { | ||
322 | 80 | 10 | if (*ptr == 0) { | |
323 | ##### | 11 | goto ret; | |
324 | - | } | ||
325 | - | |||
326 | 80 | 12-14 | if (isalnum(*ptr) || *ptr == '-') { | |
327 | - | /* legal key character */ | ||
328 | 67 | 15 | NEXT(S_KEY); | |
329 | - | } | ||
330 | - | |||
331 | 13 | 16,17 | if (*ptr == ' ' || *ptr == '\t') { | |
332 | - | /* optional whitespace before separator */ | ||
333 | 2 | 18 | *ptr = 0; | |
334 | 2 | 18 | NEXT(S_KEY_WS); | |
335 | - | } | ||
336 | - | |||
337 | 11 | 19 | if (strchr(TRAILER_SEPARATORS, *ptr)) { | |
338 | 11 | 20 | *ptr = 0; | |
339 | 11 | 20 | NEXT(S_SEP_WS); | |
340 | - | } | ||
341 | - | |||
342 | - | /* illegal character */ | ||
343 | ##### | 21 | GOTO(S_IGNORE); | |
344 | - | } | ||
345 | - | case S_KEY_WS: { | ||
346 | 4 | 22 | if (*ptr == 0) { | |
347 | ##### | 23 | goto ret; | |
348 | - | } | ||
349 | - | |||
350 | 4 | 24,25 | if (*ptr == ' ' || *ptr == '\t') { | |
351 | 2 | 26 | NEXT(S_KEY_WS); | |
352 | - | } | ||
353 | - | |||
354 | 2 | 27 | if (strchr(TRAILER_SEPARATORS, *ptr)) { | |
355 | 1 | 28 | NEXT(S_SEP_WS); | |
356 | - | } | ||
357 | - | |||
358 | - | /* illegal character */ | ||
359 | 1 | 29 | GOTO(S_IGNORE); | |
360 | - | } | ||
361 | - | case S_SEP_WS: { | ||
362 | 25 | 30 | if (*ptr == 0) { | |
363 | ##### | 31 | goto ret; | |
364 | - | } | ||
365 | - | |||
366 | 25 | 32,33 | if (*ptr == ' ' || *ptr == '\t') { | |
367 | 13 | 34 | NEXT(S_SEP_WS); | |
368 | - | } | ||
369 | - | |||
370 | 12 | 35 | value = ptr; | |
371 | 12 | 35 | NEXT(S_VALUE); | |
372 | - | } | ||
373 | - | case S_VALUE: { | ||
374 | 83 | 36 | if (*ptr == 0) { | |
375 | 1 | 37 | GOTO(S_VALUE_END); | |
376 | - | } | ||
377 | - | |||
378 | 82 | 38 | if (*ptr == '\n') { | |
379 | 13 | 39 | NEXT(S_VALUE_NL); | |
380 | - | } | ||
381 | - | |||
382 | 69 | 40 | NEXT(S_VALUE); | |
383 | - | } | ||
384 | - | case S_VALUE_NL: { | ||
385 | 13 | 41 | if (*ptr == ' ') { | |
386 | - | /* continuation; */ | ||
387 | 2 | 42 | NEXT(S_VALUE); | |
388 | - | } | ||
389 | - | |||
390 | 11 | 43 | ptr[-1] = 0; | |
391 | 11 | 43 | GOTO(S_VALUE_END); | |
392 | - | } | ||
393 | - | case S_VALUE_END: { | ||
394 | 12 | 44-49 | git_message_trailer *t = git_array_alloc(arr); | |
395 | - | |||
396 | 12 | 50 | t->key = key; | |
397 | 12 | 50 | t->value = value; | |
398 | - | |||
399 | 12 | 50 | key = NULL; | |
400 | 12 | 50 | value = NULL; | |
401 | - | |||
402 | 12 | 50 | GOTO(S_START); | |
403 | - | } | ||
404 | - | case S_IGNORE: { | ||
405 | 10 | 51 | if (*ptr == 0) { | |
406 | ##### | 52 | goto ret; | |
407 | - | } | ||
408 | - | |||
409 | 10 | 53 | if (*ptr == '\n') { | |
410 | 1 | 54 | NEXT(S_START); | |
411 | - | } | ||
412 | - | |||
413 | 9 | 55 | NEXT(S_IGNORE); | |
414 | - | } | ||
415 | - | } | ||
416 | 240 | 56 | } | |
417 | - | |||
418 | - | ret: | ||
419 | 9 | 57 | trailer_arr->_trailer_block = trailer; | |
420 | 9 | 57 | trailer_arr->trailers = arr.ptr; | |
421 | 9 | 57 | trailer_arr->count = arr.size; | |
422 | - | |||
423 | 9 | 57 | return rc; | |
424 | - | } | ||
425 | - | |||
426 | 9 | 2 | void git_message_trailer_array_free(git_message_trailer_array *arr) | |
427 | - | { | ||
428 | 9 | 2 | git__free(arr->_trailer_block); | |
429 | 9 | 3 | git__free(arr->trailers); | |
430 | 9 | 4 | } |