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