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