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