Line | Count | Source |
1 | | /* |
2 | | * Whitespace rules |
3 | | * |
4 | | * Copyright (c) 2007 Junio C Hamano |
5 | | */ |
6 | | |
7 | | #define DISABLE_SIGN_COMPARE_WARNINGS |
8 | | |
9 | | #include "git-compat-util.h" |
10 | | #include "attr.h" |
11 | | #include "strbuf.h" |
12 | | #include "ws.h" |
13 | | |
14 | | unsigned whitespace_rule_cfg = WS_DEFAULT_RULE; |
15 | | |
16 | | static struct whitespace_rule { |
17 | | const char *rule_name; |
18 | | unsigned rule_bits; |
19 | | unsigned loosens_error:1, |
20 | | exclude_default:1; |
21 | | } whitespace_rule_names[] = { |
22 | | { "trailing-space", WS_TRAILING_SPACE, 0 }, |
23 | | { "space-before-tab", WS_SPACE_BEFORE_TAB, 0 }, |
24 | | { "indent-with-non-tab", WS_INDENT_WITH_NON_TAB, 0 }, |
25 | | { "cr-at-eol", WS_CR_AT_EOL, 1 }, |
26 | | { "blank-at-eol", WS_BLANK_AT_EOL, 0 }, |
27 | | { "blank-at-eof", WS_BLANK_AT_EOF, 0 }, |
28 | | { "tab-in-indent", WS_TAB_IN_INDENT, 0, 1 }, |
29 | | { "incomplete-line", WS_INCOMPLETE_LINE, 0, 0 }, |
30 | | }; |
31 | | |
32 | | unsigned parse_whitespace_rule(const char *string) |
33 | 0 | { |
34 | 0 | unsigned rule = WS_DEFAULT_RULE; |
35 | |
|
36 | 0 | while (string) { |
37 | 0 | int i; |
38 | 0 | size_t len; |
39 | 0 | const char *ep; |
40 | 0 | const char *arg; |
41 | 0 | int negated = 0; |
42 | |
|
43 | 0 | string = string + strspn(string, ", \t\n\r"); |
44 | 0 | ep = strchrnul(string, ','); |
45 | 0 | len = ep - string; |
46 | |
|
47 | 0 | if (*string == '-') { |
48 | 0 | negated = 1; |
49 | 0 | string++; |
50 | 0 | len--; |
51 | 0 | } |
52 | 0 | if (!len) |
53 | 0 | break; |
54 | 0 | for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++) { |
55 | 0 | if (strncmp(whitespace_rule_names[i].rule_name, |
56 | 0 | string, len)) |
57 | 0 | continue; |
58 | 0 | if (negated) |
59 | 0 | rule &= ~whitespace_rule_names[i].rule_bits; |
60 | 0 | else |
61 | 0 | rule |= whitespace_rule_names[i].rule_bits; |
62 | 0 | break; |
63 | 0 | } |
64 | 0 | if (skip_prefix(string, "tabwidth=", &arg)) { |
65 | 0 | unsigned tabwidth = atoi(arg); |
66 | 0 | if (0 < tabwidth && tabwidth < 0100) { |
67 | 0 | rule &= ~WS_TAB_WIDTH_MASK; |
68 | 0 | rule |= tabwidth; |
69 | 0 | } |
70 | 0 | else |
71 | 0 | warning("tabwidth %.*s out of range", |
72 | 0 | (int)(ep - arg), arg); |
73 | 0 | } |
74 | 0 | string = ep; |
75 | 0 | } |
76 | |
|
77 | 0 | if (rule & WS_TAB_IN_INDENT && rule & WS_INDENT_WITH_NON_TAB) |
78 | 0 | die("cannot enforce both tab-in-indent and indent-with-non-tab"); |
79 | 0 | return rule; |
80 | 0 | } |
81 | | |
82 | | unsigned whitespace_rule(struct index_state *istate, const char *pathname) |
83 | 0 | { |
84 | 0 | static struct attr_check *attr_whitespace_rule; |
85 | 0 | const char *value; |
86 | |
|
87 | 0 | if (!attr_whitespace_rule) |
88 | 0 | attr_whitespace_rule = attr_check_initl("whitespace", NULL); |
89 | |
|
90 | 0 | git_check_attr(istate, pathname, attr_whitespace_rule); |
91 | 0 | value = attr_whitespace_rule->items[0].value; |
92 | 0 | if (ATTR_TRUE(value)) { |
93 | | /* true (whitespace) */ |
94 | 0 | unsigned all_rule = ws_tab_width(whitespace_rule_cfg); |
95 | 0 | int i; |
96 | 0 | for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++) |
97 | 0 | if (!whitespace_rule_names[i].loosens_error && |
98 | 0 | !whitespace_rule_names[i].exclude_default) |
99 | 0 | all_rule |= whitespace_rule_names[i].rule_bits; |
100 | 0 | return all_rule; |
101 | 0 | } else if (ATTR_FALSE(value)) { |
102 | | /* false (-whitespace) */ |
103 | 0 | return ws_tab_width(whitespace_rule_cfg); |
104 | 0 | } else if (ATTR_UNSET(value)) { |
105 | | /* reset to default (!whitespace) */ |
106 | 0 | return whitespace_rule_cfg; |
107 | 0 | } else { |
108 | | /* string */ |
109 | 0 | return parse_whitespace_rule(value); |
110 | 0 | } |
111 | 0 | } |
112 | | |
113 | | /* The returned string should be freed by the caller. */ |
114 | | char *whitespace_error_string(unsigned ws) |
115 | 0 | { |
116 | 0 | struct strbuf err = STRBUF_INIT; |
117 | 0 | if ((ws & WS_TRAILING_SPACE) == WS_TRAILING_SPACE) |
118 | 0 | strbuf_addstr(&err, "trailing whitespace"); |
119 | 0 | else { |
120 | 0 | if (ws & WS_BLANK_AT_EOL) |
121 | 0 | strbuf_addstr(&err, "trailing whitespace"); |
122 | 0 | if (ws & WS_BLANK_AT_EOF) { |
123 | 0 | if (err.len) |
124 | 0 | strbuf_addstr(&err, ", "); |
125 | 0 | strbuf_addstr(&err, "new blank line at EOF"); |
126 | 0 | } |
127 | 0 | } |
128 | 0 | if (ws & WS_SPACE_BEFORE_TAB) { |
129 | 0 | if (err.len) |
130 | 0 | strbuf_addstr(&err, ", "); |
131 | 0 | strbuf_addstr(&err, "space before tab in indent"); |
132 | 0 | } |
133 | 0 | if (ws & WS_INDENT_WITH_NON_TAB) { |
134 | 0 | if (err.len) |
135 | 0 | strbuf_addstr(&err, ", "); |
136 | 0 | strbuf_addstr(&err, "indent with spaces"); |
137 | 0 | } |
138 | 0 | if (ws & WS_TAB_IN_INDENT) { |
139 | 0 | if (err.len) |
140 | 0 | strbuf_addstr(&err, ", "); |
141 | 0 | strbuf_addstr(&err, "tab in indent"); |
142 | 0 | } |
143 | 0 | if (ws & WS_INCOMPLETE_LINE) { |
144 | 0 | if (err.len) |
145 | 0 | strbuf_addstr(&err, ", "); |
146 | 0 | strbuf_addstr(&err, "no newline at the end of file"); |
147 | 0 | } |
148 | 0 | return strbuf_detach(&err, NULL); |
149 | 0 | } |
150 | | |
151 | | /* If stream is non-NULL, emits the line after checking. */ |
152 | | static unsigned ws_check_emit_1(const char *line, int len, unsigned ws_rule, |
153 | | FILE *stream, const char *set, |
154 | | const char *reset, const char *ws) |
155 | 0 | { |
156 | 0 | unsigned result = 0; |
157 | 0 | int written = 0; |
158 | 0 | int trailing_whitespace = -1; |
159 | 0 | int trailing_newline = 0; |
160 | 0 | int trailing_carriage_return = 0; |
161 | 0 | int i; |
162 | | |
163 | | /* Logic is simpler if we temporarily ignore the trailing newline. */ |
164 | 0 | if (len > 0 && line[len - 1] == '\n') { |
165 | 0 | trailing_newline = 1; |
166 | 0 | len--; |
167 | 0 | } |
168 | 0 | if ((ws_rule & WS_CR_AT_EOL) && |
169 | 0 | len > 0 && line[len - 1] == '\r') { |
170 | 0 | trailing_carriage_return = 1; |
171 | 0 | len--; |
172 | 0 | } |
173 | | |
174 | | /* Check for trailing whitespace. */ |
175 | 0 | if (ws_rule & WS_BLANK_AT_EOL) { |
176 | 0 | for (i = len - 1; i >= 0; i--) { |
177 | 0 | if (isspace(line[i])) { |
178 | 0 | trailing_whitespace = i; |
179 | 0 | result |= WS_BLANK_AT_EOL; |
180 | 0 | } |
181 | 0 | else |
182 | 0 | break; |
183 | 0 | } |
184 | 0 | } |
185 | |
|
186 | 0 | if (trailing_whitespace == -1) |
187 | 0 | trailing_whitespace = len; |
188 | |
|
189 | 0 | if (!trailing_newline && (ws_rule & WS_INCOMPLETE_LINE)) |
190 | 0 | result |= WS_INCOMPLETE_LINE; |
191 | | |
192 | | /* Check indentation */ |
193 | 0 | for (i = 0; i < trailing_whitespace; i++) { |
194 | 0 | if (line[i] == ' ') |
195 | 0 | continue; |
196 | 0 | if (line[i] != '\t') |
197 | 0 | break; |
198 | 0 | if ((ws_rule & WS_SPACE_BEFORE_TAB) && written < i) { |
199 | 0 | result |= WS_SPACE_BEFORE_TAB; |
200 | 0 | if (stream) { |
201 | 0 | fputs(ws, stream); |
202 | 0 | fwrite(line + written, i - written, 1, stream); |
203 | 0 | fputs(reset, stream); |
204 | 0 | fwrite(line + i, 1, 1, stream); |
205 | 0 | } |
206 | 0 | } else if (ws_rule & WS_TAB_IN_INDENT) { |
207 | 0 | result |= WS_TAB_IN_INDENT; |
208 | 0 | if (stream) { |
209 | 0 | fwrite(line + written, i - written, 1, stream); |
210 | 0 | fputs(ws, stream); |
211 | 0 | fwrite(line + i, 1, 1, stream); |
212 | 0 | fputs(reset, stream); |
213 | 0 | } |
214 | 0 | } else if (stream) { |
215 | 0 | fwrite(line + written, i - written + 1, 1, stream); |
216 | 0 | } |
217 | 0 | written = i + 1; |
218 | 0 | } |
219 | | |
220 | | /* Check for indent using non-tab. */ |
221 | 0 | if ((ws_rule & WS_INDENT_WITH_NON_TAB) && i - written >= ws_tab_width(ws_rule)) { |
222 | 0 | result |= WS_INDENT_WITH_NON_TAB; |
223 | 0 | if (stream) { |
224 | 0 | fputs(ws, stream); |
225 | 0 | fwrite(line + written, i - written, 1, stream); |
226 | 0 | fputs(reset, stream); |
227 | 0 | } |
228 | 0 | written = i; |
229 | 0 | } |
230 | |
|
231 | 0 | if (stream) { |
232 | | /* |
233 | | * Now the rest of the line starts at "written". |
234 | | * The non-highlighted part ends at "trailing_whitespace". |
235 | | */ |
236 | | |
237 | | /* Emit non-highlighted (middle) segment. */ |
238 | 0 | if (trailing_whitespace - written > 0) { |
239 | 0 | fputs(set, stream); |
240 | 0 | fwrite(line + written, |
241 | 0 | trailing_whitespace - written, 1, stream); |
242 | 0 | fputs(reset, stream); |
243 | 0 | } |
244 | | |
245 | | /* Highlight errors in trailing whitespace. */ |
246 | 0 | if (trailing_whitespace != len) { |
247 | 0 | fputs(ws, stream); |
248 | 0 | fwrite(line + trailing_whitespace, |
249 | 0 | len - trailing_whitespace, 1, stream); |
250 | 0 | fputs(reset, stream); |
251 | 0 | } |
252 | 0 | if (trailing_carriage_return) |
253 | 0 | fputc('\r', stream); |
254 | 0 | if (trailing_newline) |
255 | 0 | fputc('\n', stream); |
256 | 0 | } |
257 | 0 | return result; |
258 | 0 | } |
259 | | |
260 | | void ws_check_emit(const char *line, int len, unsigned ws_rule, |
261 | | FILE *stream, const char *set, |
262 | | const char *reset, const char *ws) |
263 | 0 | { |
264 | 0 | (void)ws_check_emit_1(line, len, ws_rule, stream, set, reset, ws); |
265 | 0 | } |
266 | | |
267 | | unsigned ws_check(const char *line, int len, unsigned ws_rule) |
268 | 0 | { |
269 | 0 | return ws_check_emit_1(line, len, ws_rule, NULL, NULL, NULL, NULL); |
270 | 0 | } |
271 | | |
272 | | int ws_blank_line(const char *line, int len) |
273 | 0 | { |
274 | | /* |
275 | | * We _might_ want to treat CR differently from other |
276 | | * whitespace characters when ws_rule has WS_CR_AT_EOL, but |
277 | | * for now we just use this stupid definition. |
278 | | */ |
279 | 0 | while (len-- > 0) { |
280 | 0 | if (!isspace(*line)) |
281 | 0 | return 0; |
282 | 0 | line++; |
283 | 0 | } |
284 | 0 | return 1; |
285 | 0 | } |
286 | | |
287 | | /* Copy the line onto the end of the strbuf while fixing whitespaces */ |
288 | | void ws_fix_copy(struct strbuf *dst, const char *src, int len, unsigned ws_rule, int *error_count) |
289 | 0 | { |
290 | | /* |
291 | | * len is number of bytes to be copied from src, starting |
292 | | * at src. Typically src[len-1] is '\n', unless this is |
293 | | * the incomplete last line. |
294 | | */ |
295 | 0 | int i; |
296 | 0 | int add_nl_to_tail = 0; |
297 | 0 | int add_cr_to_tail = 0; |
298 | 0 | int fixed = 0; |
299 | 0 | int last_tab_in_indent = -1; |
300 | 0 | int last_space_in_indent = -1; |
301 | 0 | int need_fix_leading_space = 0; |
302 | | |
303 | | /* |
304 | | * Remembering that we need to add '\n' at the end |
305 | | * is sufficient to fix an incomplete line. |
306 | | */ |
307 | 0 | if (ws_rule & WS_INCOMPLETE_LINE) { |
308 | 0 | if (0 < len && src[len - 1] != '\n') { |
309 | 0 | fixed = 1; |
310 | 0 | add_nl_to_tail = 1; |
311 | 0 | } |
312 | 0 | } |
313 | | |
314 | | /* |
315 | | * Strip trailing whitespace |
316 | | */ |
317 | 0 | if (ws_rule & WS_BLANK_AT_EOL) { |
318 | 0 | if (0 < len && src[len - 1] == '\n') { |
319 | 0 | add_nl_to_tail = 1; |
320 | 0 | len--; |
321 | 0 | if (0 < len && src[len - 1] == '\r') { |
322 | 0 | add_cr_to_tail = !!(ws_rule & WS_CR_AT_EOL); |
323 | 0 | len--; |
324 | 0 | } |
325 | 0 | } |
326 | 0 | if (0 < len && isspace(src[len - 1])) { |
327 | 0 | while (0 < len && isspace(src[len-1])) |
328 | 0 | len--; |
329 | 0 | fixed = 1; |
330 | 0 | } |
331 | 0 | } |
332 | | |
333 | | /* |
334 | | * Check leading whitespaces (indent) |
335 | | */ |
336 | 0 | for (i = 0; i < len; i++) { |
337 | 0 | char ch = src[i]; |
338 | 0 | if (ch == '\t') { |
339 | 0 | last_tab_in_indent = i; |
340 | 0 | if ((ws_rule & WS_SPACE_BEFORE_TAB) && |
341 | 0 | 0 <= last_space_in_indent) |
342 | 0 | need_fix_leading_space = 1; |
343 | 0 | } else if (ch == ' ') { |
344 | 0 | last_space_in_indent = i; |
345 | 0 | if ((ws_rule & WS_INDENT_WITH_NON_TAB) && |
346 | 0 | ws_tab_width(ws_rule) <= i - last_tab_in_indent) |
347 | 0 | need_fix_leading_space = 1; |
348 | 0 | } else |
349 | 0 | break; |
350 | 0 | } |
351 | |
|
352 | 0 | if (need_fix_leading_space) { |
353 | | /* Process indent ourselves */ |
354 | 0 | int consecutive_spaces = 0; |
355 | 0 | int last = last_tab_in_indent + 1; |
356 | |
|
357 | 0 | if (ws_rule & WS_INDENT_WITH_NON_TAB) { |
358 | | /* have "last" point at one past the indent */ |
359 | 0 | if (last_tab_in_indent < last_space_in_indent) |
360 | 0 | last = last_space_in_indent + 1; |
361 | 0 | else |
362 | 0 | last = last_tab_in_indent + 1; |
363 | 0 | } |
364 | | |
365 | | /* |
366 | | * between src[0..last-1], strip the funny spaces, |
367 | | * updating them to tab as needed. |
368 | | */ |
369 | 0 | for (i = 0; i < last; i++) { |
370 | 0 | char ch = src[i]; |
371 | 0 | if (ch != ' ') { |
372 | 0 | consecutive_spaces = 0; |
373 | 0 | strbuf_addch(dst, ch); |
374 | 0 | } else { |
375 | 0 | consecutive_spaces++; |
376 | 0 | if (consecutive_spaces == ws_tab_width(ws_rule)) { |
377 | 0 | strbuf_addch(dst, '\t'); |
378 | 0 | consecutive_spaces = 0; |
379 | 0 | } |
380 | 0 | } |
381 | 0 | } |
382 | 0 | while (0 < consecutive_spaces--) |
383 | 0 | strbuf_addch(dst, ' '); |
384 | 0 | len -= last; |
385 | 0 | src += last; |
386 | 0 | fixed = 1; |
387 | 0 | } else if ((ws_rule & WS_TAB_IN_INDENT) && last_tab_in_indent >= 0) { |
388 | | /* Expand tabs into spaces */ |
389 | 0 | int start = dst->len; |
390 | 0 | int last = last_tab_in_indent + 1; |
391 | 0 | for (i = 0; i < last; i++) { |
392 | 0 | if (src[i] == '\t') |
393 | 0 | do { |
394 | 0 | strbuf_addch(dst, ' '); |
395 | 0 | } while ((dst->len - start) % ws_tab_width(ws_rule)); |
396 | 0 | else |
397 | 0 | strbuf_addch(dst, src[i]); |
398 | 0 | } |
399 | 0 | len -= last; |
400 | 0 | src += last; |
401 | 0 | fixed = 1; |
402 | 0 | } |
403 | |
|
404 | 0 | strbuf_add(dst, src, len); |
405 | 0 | if (add_cr_to_tail) |
406 | 0 | strbuf_addch(dst, '\r'); |
407 | 0 | if (add_nl_to_tail) |
408 | 0 | strbuf_addch(dst, '\n'); |
409 | 0 | if (fixed && error_count) |
410 | 0 | (*error_count)++; |
411 | 0 | } |