blob: 9456e2fdbe3766b89497cf740b6280770daf5fd8 [file] [log] [blame]
Junio C Hamanocf1b7862007-12-06 00:14:14 -08001/*
2 * Whitespace rules
3 *
4 * Copyright (c) 2007 Junio C Hamano
5 */
Elijah Newren64122312023-04-22 20:17:16 +00006#include "git-compat-util.h"
Junio C Hamanocf1b7862007-12-06 00:14:14 -08007#include "attr.h"
Elijah Newren69a63fe2023-04-22 20:17:08 +00008#include "strbuf.h"
Elijah Newren64122312023-04-22 20:17:16 +00009#include "ws.h"
10
11unsigned whitespace_rule_cfg = WS_DEFAULT_RULE;
Junio C Hamanocf1b7862007-12-06 00:14:14 -080012
13static struct whitespace_rule {
14 const char *rule_name;
15 unsigned rule_bits;
Junio C Hamano727c3712010-04-03 00:37:00 +010016 unsigned loosens_error:1,
17 exclude_default:1;
Junio C Hamanocf1b7862007-12-06 00:14:14 -080018} whitespace_rule_names[] = {
Junio C Hamanoa4379002009-06-21 02:35:18 -070019 { "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 },
Junio C Hamanoafd9db42009-09-15 03:28:08 -070023 { "blank-at-eol", WS_BLANK_AT_EOL, 0 },
24 { "blank-at-eof", WS_BLANK_AT_EOF, 0 },
Chris Webb3e3ec2a2010-04-03 00:37:08 +010025 { "tab-in-indent", WS_TAB_IN_INDENT, 0, 1 },
Junio C Hamanocf1b7862007-12-06 00:14:14 -080026};
27
28unsigned parse_whitespace_rule(const char *string)
29{
30 unsigned rule = WS_DEFAULT_RULE;
31
32 while (string) {
33 int i;
34 size_t len;
35 const char *ep;
Jeff Kingd43b9932023-01-07 08:26:44 -050036 const char *arg;
Junio C Hamanocf1b7862007-12-06 00:14:14 -080037 int negated = 0;
38
39 string = string + strspn(string, ", \t\n\r");
Rohit Mani2c5495f2014-03-07 22:48:31 -080040 ep = strchrnul(string, ',');
41 len = ep - string;
Junio C Hamanocf1b7862007-12-06 00:14:14 -080042
43 if (*string == '-') {
44 negated = 1;
45 string++;
46 len--;
47 }
48 if (!len)
49 break;
50 for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++) {
51 if (strncmp(whitespace_rule_names[i].rule_name,
52 string, len))
53 continue;
54 if (negated)
55 rule &= ~whitespace_rule_names[i].rule_bits;
56 else
57 rule |= whitespace_rule_names[i].rule_bits;
58 break;
59 }
Jeff Kingd43b9932023-01-07 08:26:44 -050060 if (skip_prefix(string, "tabwidth=", &arg)) {
61 unsigned tabwidth = atoi(arg);
Johannes Sixtf4b05a42010-11-30 09:29:11 +010062 if (0 < tabwidth && tabwidth < 0100) {
63 rule &= ~WS_TAB_WIDTH_MASK;
64 rule |= tabwidth;
65 }
66 else
67 warning("tabwidth %.*s out of range",
Jeff Kingd43b9932023-01-07 08:26:44 -050068 (int)(ep - arg), arg);
Johannes Sixtf4b05a42010-11-30 09:29:11 +010069 }
Junio C Hamanocf1b7862007-12-06 00:14:14 -080070 string = ep;
71 }
Chris Webb3e3ec2a2010-04-03 00:37:08 +010072
73 if (rule & WS_TAB_IN_INDENT && rule & WS_INDENT_WITH_NON_TAB)
74 die("cannot enforce both tab-in-indent and indent-with-non-tab");
Junio C Hamanocf1b7862007-12-06 00:14:14 -080075 return rule;
76}
77
Nguyễn Thái Ngọc Duy26d024e2018-09-21 17:57:37 +020078unsigned whitespace_rule(struct index_state *istate, const char *pathname)
Junio C Hamanocf1b7862007-12-06 00:14:14 -080079{
Junio C Hamano2aef63d2017-01-27 18:01:57 -080080 static struct attr_check *attr_whitespace_rule;
Torsten Bögershausend64324c2018-09-12 21:32:02 +020081 const char *value;
Junio C Hamanocf1b7862007-12-06 00:14:14 -080082
Junio C Hamano2aef63d2017-01-27 18:01:57 -080083 if (!attr_whitespace_rule)
84 attr_whitespace_rule = attr_check_initl("whitespace", NULL);
85
John Cai44451a22023-05-06 04:15:29 +000086 git_check_attr(istate, pathname, attr_whitespace_rule);
Torsten Bögershausend64324c2018-09-12 21:32:02 +020087 value = attr_whitespace_rule->items[0].value;
88 if (ATTR_TRUE(value)) {
89 /* true (whitespace) */
90 unsigned all_rule = ws_tab_width(whitespace_rule_cfg);
91 int i;
92 for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++)
93 if (!whitespace_rule_names[i].loosens_error &&
94 !whitespace_rule_names[i].exclude_default)
95 all_rule |= whitespace_rule_names[i].rule_bits;
96 return all_rule;
97 } else if (ATTR_FALSE(value)) {
98 /* false (-whitespace) */
99 return ws_tab_width(whitespace_rule_cfg);
100 } else if (ATTR_UNSET(value)) {
101 /* reset to default (!whitespace) */
Junio C Hamanocf1b7862007-12-06 00:14:14 -0800102 return whitespace_rule_cfg;
Torsten Bögershausend64324c2018-09-12 21:32:02 +0200103 } else {
104 /* string */
105 return parse_whitespace_rule(value);
Junio C Hamanocf1b7862007-12-06 00:14:14 -0800106 }
107}
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100108
109/* The returned string should be freed by the caller. */
110char *whitespace_error_string(unsigned ws)
111{
Brandon Caseyf285a2d2008-10-09 14:12:12 -0500112 struct strbuf err = STRBUF_INIT;
Junio C Hamanoaeb84b02009-09-05 22:21:17 -0700113 if ((ws & WS_TRAILING_SPACE) == WS_TRAILING_SPACE)
Wincent Colaiuta420f4f02007-12-14 12:23:43 +0100114 strbuf_addstr(&err, "trailing whitespace");
Junio C Hamanoaeb84b02009-09-05 22:21:17 -0700115 else {
116 if (ws & WS_BLANK_AT_EOL)
117 strbuf_addstr(&err, "trailing whitespace");
118 if (ws & WS_BLANK_AT_EOF) {
119 if (err.len)
120 strbuf_addstr(&err, ", ");
121 strbuf_addstr(&err, "new blank line at EOF");
122 }
123 }
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100124 if (ws & WS_SPACE_BEFORE_TAB) {
125 if (err.len)
126 strbuf_addstr(&err, ", ");
Wincent Colaiuta420f4f02007-12-14 12:23:43 +0100127 strbuf_addstr(&err, "space before tab in indent");
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100128 }
129 if (ws & WS_INDENT_WITH_NON_TAB) {
130 if (err.len)
131 strbuf_addstr(&err, ", ");
Wincent Colaiuta420f4f02007-12-14 12:23:43 +0100132 strbuf_addstr(&err, "indent with spaces");
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100133 }
Chris Webb3e3ec2a2010-04-03 00:37:08 +0100134 if (ws & WS_TAB_IN_INDENT) {
135 if (err.len)
136 strbuf_addstr(&err, ", ");
137 strbuf_addstr(&err, "tab in indent");
138 }
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100139 return strbuf_detach(&err, NULL);
140}
141
142/* If stream is non-NULL, emits the line after checking. */
Junio C Hamano8f8841e2008-06-26 15:35:21 -0700143static 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)
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100146{
147 unsigned result = 0;
J. Bruce Fields954ecd42007-12-16 11:31:39 -0500148 int written = 0;
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100149 int trailing_whitespace = -1;
150 int trailing_newline = 0;
Junio C Hamanob2979ff2008-01-15 00:59:05 -0800151 int trailing_carriage_return = 0;
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100152 int i;
153
154 /* Logic is simpler if we temporarily ignore the trailing newline. */
155 if (len > 0 && line[len - 1] == '\n') {
156 trailing_newline = 1;
157 len--;
158 }
Junio C Hamanob2979ff2008-01-15 00:59:05 -0800159 if ((ws_rule & WS_CR_AT_EOL) &&
160 len > 0 && line[len - 1] == '\r') {
161 trailing_carriage_return = 1;
162 len--;
163 }
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100164
165 /* Check for trailing whitespace. */
Junio C Hamanoaeb84b02009-09-05 22:21:17 -0700166 if (ws_rule & WS_BLANK_AT_EOL) {
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100167 for (i = len - 1; i >= 0; i--) {
168 if (isspace(line[i])) {
169 trailing_whitespace = i;
Junio C Hamanoaeb84b02009-09-05 22:21:17 -0700170 result |= WS_BLANK_AT_EOL;
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100171 }
172 else
173 break;
174 }
175 }
176
Kevin Ballardcfd1a982010-10-20 15:17:26 -0700177 if (trailing_whitespace == -1)
178 trailing_whitespace = len;
179
Chris Webb3e3ec2a2010-04-03 00:37:08 +0100180 /* Check indentation */
Kevin Ballardcfd1a982010-10-20 15:17:26 -0700181 for (i = 0; i < trailing_whitespace; i++) {
J. Bruce Fields9afa2d42007-12-16 11:31:40 -0500182 if (line[i] == ' ')
J. Bruce Fields10209992007-12-16 11:31:38 -0500183 continue;
J. Bruce Fields10209992007-12-16 11:31:38 -0500184 if (line[i] != '\t')
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100185 break;
J. Bruce Fieldsffe56882007-12-16 11:31:41 -0500186 if ((ws_rule & WS_SPACE_BEFORE_TAB) && written < i) {
J. Bruce Fields10209992007-12-16 11:31:38 -0500187 result |= WS_SPACE_BEFORE_TAB;
J. Bruce Fieldsffe56882007-12-16 11:31:41 -0500188 if (stream) {
189 fputs(ws, stream);
190 fwrite(line + written, i - written, 1, stream);
191 fputs(reset, stream);
Chris Webb3e3ec2a2010-04-03 00:37:08 +0100192 fwrite(line + i, 1, 1, stream);
J. Bruce Fieldsffe56882007-12-16 11:31:41 -0500193 }
Chris Webb3e3ec2a2010-04-03 00:37:08 +0100194 } else if (ws_rule & WS_TAB_IN_INDENT) {
195 result |= WS_TAB_IN_INDENT;
196 if (stream) {
197 fwrite(line + written, i - written, 1, stream);
198 fputs(ws, stream);
199 fwrite(line + i, 1, 1, stream);
200 fputs(reset, stream);
201 }
202 } else if (stream) {
203 fwrite(line + written, i - written + 1, 1, stream);
204 }
J. Bruce Fields9afa2d42007-12-16 11:31:40 -0500205 written = i + 1;
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100206 }
207
208 /* Check for indent using non-tab. */
Johannes Sixtf4b05a42010-11-30 09:29:11 +0100209 if ((ws_rule & WS_INDENT_WITH_NON_TAB) && i - written >= ws_tab_width(ws_rule)) {
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100210 result |= WS_INDENT_WITH_NON_TAB;
J. Bruce Fieldsffe56882007-12-16 11:31:41 -0500211 if (stream) {
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100212 fputs(ws, stream);
J. Bruce Fieldsffe56882007-12-16 11:31:41 -0500213 fwrite(line + written, i - written, 1, stream);
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100214 fputs(reset, stream);
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100215 }
J. Bruce Fieldsffe56882007-12-16 11:31:41 -0500216 written = i;
217 }
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100218
J. Bruce Fieldsffe56882007-12-16 11:31:41 -0500219 if (stream) {
Junio C Hamanob2979ff2008-01-15 00:59:05 -0800220 /*
221 * Now the rest of the line starts at "written".
222 * The non-highlighted part ends at "trailing_whitespace".
223 */
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100224
225 /* Emit non-highlighted (middle) segment. */
J. Bruce Fields954ecd42007-12-16 11:31:39 -0500226 if (trailing_whitespace - written > 0) {
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100227 fputs(set, stream);
J. Bruce Fields954ecd42007-12-16 11:31:39 -0500228 fwrite(line + written,
229 trailing_whitespace - written, 1, stream);
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100230 fputs(reset, stream);
231 }
232
233 /* Highlight errors in trailing whitespace. */
234 if (trailing_whitespace != len) {
235 fputs(ws, stream);
236 fwrite(line + trailing_whitespace,
237 len - trailing_whitespace, 1, stream);
238 fputs(reset, stream);
239 }
Junio C Hamanob2979ff2008-01-15 00:59:05 -0800240 if (trailing_carriage_return)
241 fputc('\r', stream);
Wincent Colaiutac1795bb2007-12-13 14:32:29 +0100242 if (trailing_newline)
243 fputc('\n', stream);
244 }
245 return result;
246}
Junio C Hamanofe3403c2008-02-23 16:59:16 -0800247
Junio C Hamano8f8841e2008-06-26 15:35:21 -0700248void 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{
252 (void)ws_check_emit_1(line, len, ws_rule, stream, set, reset, ws);
253}
254
255unsigned ws_check(const char *line, int len, unsigned ws_rule)
256{
257 return ws_check_emit_1(line, len, ws_rule, NULL, NULL, NULL, NULL);
258}
259
Jeff Kingc5224f02022-12-13 06:12:58 -0500260int ws_blank_line(const char *line, int len)
Junio C Hamano877f23c2008-06-26 15:36:59 -0700261{
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 while (len-- > 0) {
268 if (!isspace(*line))
269 return 0;
270 line++;
271 }
272 return 1;
273}
274
Chris Webbd511bd32010-04-03 00:37:23 +0100275/* Copy the line onto the end of the strbuf while fixing whitespaces */
276void ws_fix_copy(struct strbuf *dst, const char *src, int len, unsigned ws_rule, int *error_count)
Junio C Hamanofe3403c2008-02-23 16:59:16 -0800277{
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 int i;
284 int add_nl_to_tail = 0;
285 int add_cr_to_tail = 0;
286 int fixed = 0;
287 int last_tab_in_indent = -1;
288 int last_space_in_indent = -1;
289 int need_fix_leading_space = 0;
Junio C Hamanofe3403c2008-02-23 16:59:16 -0800290
291 /*
292 * Strip trailing whitespace
293 */
Junio C Hamanoafd9db42009-09-15 03:28:08 -0700294 if (ws_rule & WS_BLANK_AT_EOL) {
Junio C Hamano422a82f2009-07-25 01:29:20 -0700295 if (0 < len && src[len - 1] == '\n') {
Junio C Hamanofe3403c2008-02-23 16:59:16 -0800296 add_nl_to_tail = 1;
297 len--;
Junio C Hamano422a82f2009-07-25 01:29:20 -0700298 if (0 < len && src[len - 1] == '\r') {
Junio C Hamanofe3403c2008-02-23 16:59:16 -0800299 add_cr_to_tail = !!(ws_rule & WS_CR_AT_EOL);
300 len--;
301 }
302 }
303 if (0 < len && isspace(src[len - 1])) {
304 while (0 < len && isspace(src[len-1]))
305 len--;
306 fixed = 1;
307 }
308 }
309
310 /*
311 * Check leading whitespaces (indent)
312 */
313 for (i = 0; i < len; i++) {
314 char ch = src[i];
315 if (ch == '\t') {
316 last_tab_in_indent = i;
317 if ((ws_rule & WS_SPACE_BEFORE_TAB) &&
318 0 <= last_space_in_indent)
319 need_fix_leading_space = 1;
320 } else if (ch == ' ') {
321 last_space_in_indent = i;
322 if ((ws_rule & WS_INDENT_WITH_NON_TAB) &&
Johannes Sixtf4b05a42010-11-30 09:29:11 +0100323 ws_tab_width(ws_rule) <= i - last_tab_in_indent)
Junio C Hamanofe3403c2008-02-23 16:59:16 -0800324 need_fix_leading_space = 1;
325 } else
326 break;
327 }
328
Junio C Hamanofe3403c2008-02-23 16:59:16 -0800329 if (need_fix_leading_space) {
330 /* Process indent ourselves */
331 int consecutive_spaces = 0;
332 int last = last_tab_in_indent + 1;
333
334 if (ws_rule & WS_INDENT_WITH_NON_TAB) {
335 /* have "last" point at one past the indent */
336 if (last_tab_in_indent < last_space_in_indent)
337 last = last_space_in_indent + 1;
338 else
339 last = last_tab_in_indent + 1;
340 }
341
342 /*
343 * between src[0..last-1], strip the funny spaces,
344 * updating them to tab as needed.
345 */
346 for (i = 0; i < last; i++) {
347 char ch = src[i];
348 if (ch != ' ') {
349 consecutive_spaces = 0;
Chris Webbd511bd32010-04-03 00:37:23 +0100350 strbuf_addch(dst, ch);
Junio C Hamanofe3403c2008-02-23 16:59:16 -0800351 } else {
352 consecutive_spaces++;
Johannes Sixtf4b05a42010-11-30 09:29:11 +0100353 if (consecutive_spaces == ws_tab_width(ws_rule)) {
Chris Webbd511bd32010-04-03 00:37:23 +0100354 strbuf_addch(dst, '\t');
Junio C Hamanofe3403c2008-02-23 16:59:16 -0800355 consecutive_spaces = 0;
356 }
357 }
358 }
359 while (0 < consecutive_spaces--)
Chris Webbd511bd32010-04-03 00:37:23 +0100360 strbuf_addch(dst, ' ');
Junio C Hamanofe3403c2008-02-23 16:59:16 -0800361 len -= last;
362 src += last;
363 fixed = 1;
Chris Webb4e35c512010-04-03 00:37:30 +0100364 } else if ((ws_rule & WS_TAB_IN_INDENT) && last_tab_in_indent >= 0) {
365 /* Expand tabs into spaces */
Johannes Sixtd35711a2010-11-30 09:22:04 +0100366 int start = dst->len;
Chris Webb4e35c512010-04-03 00:37:30 +0100367 int last = last_tab_in_indent + 1;
368 for (i = 0; i < last; i++) {
369 if (src[i] == '\t')
370 do {
371 strbuf_addch(dst, ' ');
Johannes Sixtf4b05a42010-11-30 09:29:11 +0100372 } while ((dst->len - start) % ws_tab_width(ws_rule));
Chris Webb4e35c512010-04-03 00:37:30 +0100373 else
374 strbuf_addch(dst, src[i]);
375 }
376 len -= last;
377 src += last;
378 fixed = 1;
Junio C Hamanofe3403c2008-02-23 16:59:16 -0800379 }
380
Chris Webbd511bd32010-04-03 00:37:23 +0100381 strbuf_add(dst, src, len);
Junio C Hamanofe3403c2008-02-23 16:59:16 -0800382 if (add_cr_to_tail)
Chris Webbd511bd32010-04-03 00:37:23 +0100383 strbuf_addch(dst, '\r');
Junio C Hamanofe3403c2008-02-23 16:59:16 -0800384 if (add_nl_to_tail)
Chris Webbd511bd32010-04-03 00:37:23 +0100385 strbuf_addch(dst, '\n');
Junio C Hamanofe3403c2008-02-23 16:59:16 -0800386 if (fixed && error_count)
387 (*error_count)++;
Junio C Hamanofe3403c2008-02-23 16:59:16 -0800388}