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 } whitespace_rule_names[] = {
  14         { "trailing-space", WS_TRAILING_SPACE },
  15         { "space-before-tab", WS_SPACE_BEFORE_TAB },
  16         { "indent-with-non-tab", WS_INDENT_WITH_NON_TAB },
  17         { "cr-at-eol", WS_CR_AT_EOL },
  18         { "blank-at-eol", WS_BLANK_AT_EOL },
  19         { "blank-at-eof", WS_BLANK_AT_EOF },
  20 };
  21
  22 unsigned parse_whitespace_rule(const char *string)
  23 {
  24         unsigned rule = WS_DEFAULT_RULE;
  25
  26         while (string) {
  27                 int i;
  28                 size_t len;
  29                 const char *ep;
  30                 int negated = 0;
  31
  32                 string = string + strspn(string, ", \t\n\r");
  33                 ep = strchr(string, ',');
  34                 if (!ep)
  35                         len = strlen(string);
  36                 else
  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                 string = ep;
  57         }
  58         return rule;
  59 }
  60
  61 static void setup_whitespace_attr_check(struct git_attr_check *check)
  62 {
  63         static struct git_attr *attr_whitespace;
  64
  65         if (!attr_whitespace)
  66                 attr_whitespace = git_attr("whitespace", 10);
  67         check[0].attr = attr_whitespace;
  68 }
  69
  70 unsigned whitespace_rule(const char *pathname)
  71 {
  72         struct git_attr_check attr_whitespace_rule;
  73
  74         setup_whitespace_attr_check(&attr_whitespace_rule);
  75         if (!git_checkattr(pathname, 1, &attr_whitespace_rule)) {
  76                 const char *value;
  77
  78                 value = attr_whitespace_rule.value;
  79                 if (ATTR_TRUE(value)) {
  80                         /* true (whitespace) */
  81                         unsigned all_rule = 0;
  82                         int i;
  83                         for (i = 0; i < ARRAY_SIZE(whitespace_rule_names); i++)
  84                                 all_rule |= whitespace_rule_names[i].rule_bits;
  85                         return all_rule;
  86                 } else if (ATTR_FALSE(value)) {
  87                         /* false (-whitespace) */
  88                         return 0;
  89                 } else if (ATTR_UNSET(value)) {
  90                         /* reset to default (!whitespace) */
  91                         return whitespace_rule_cfg;
  92                 } else {
  93                         /* string */
  94                         return parse_whitespace_rule(value);
  95                 }
  96         } else {
  97                 return whitespace_rule_cfg;
  98         }
  99 }
 100
 101 /* The returned string should be freed by the caller. */
 102 char *whitespace_error_string(unsigned ws)
 103 {
 104         struct strbuf err;
 105
 106         strbuf_init(&err, 0);
 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             (2 <= len && isspace(src[len-2]))) {
 277                 if (src[len - 1] == '\n') {
 278                         add_nl_to_tail = 1;
 279                         len--;
 280                         if (1 < len && src[len - 1] == '\r') {
 281                                 add_cr_to_tail = !!(ws_rule & WS_CR_AT_EOL);
 282                                 len--;
 283                         }
 284                 }
 285                 if (0 < len && isspace(src[len - 1])) {
 286                         while (0 < len && isspace(src[len-1]))
 287                                 len--;
 288                         fixed = 1;
 289                 }
 290         }
 291
 292         /*
 293          * Check leading whitespaces (indent)
 294          */
 295         for (i = 0; i < len; i++) {
 296                 char ch = src[i];
 297                 if (ch == '\t') {
 298                         last_tab_in_indent = i;
 299                         if ((ws_rule & WS_SPACE_BEFORE_TAB) &&
 300                             0 <= last_space_in_indent)
 301                             need_fix_leading_space = 1;
 302                 } else if (ch == ' ') {
 303                         last_space_in_indent = i;
 304                         if ((ws_rule & WS_INDENT_WITH_NON_TAB) &&
 305                             8 <= i - last_tab_in_indent)
 306                                 need_fix_leading_space = 1;
 307                 } else
 308                         break;
 309         }
 310
 311         buf = dst;
 312         if (need_fix_leading_space) {
 313                 /* Process indent ourselves */
 314                 int consecutive_spaces = 0;
 315                 int last = last_tab_in_indent + 1;
 316
 317                 if (ws_rule & WS_INDENT_WITH_NON_TAB) {
 318                         /* have "last" point at one past the indent */
 319                         if (last_tab_in_indent < last_space_in_indent)
 320                                 last = last_space_in_indent + 1;
 321                         else
 322                                 last = last_tab_in_indent + 1;
 323                 }
 324
 325                 /*
 326                  * between src[0..last-1], strip the funny spaces,
 327                  * updating them to tab as needed.
 328                  */
 329                 for (i = 0; i < last; i++) {
 330                         char ch = src[i];
 331                         if (ch != ' ') {
 332                                 consecutive_spaces = 0;
 333                                 *dst++ = ch;
 334                         } else {
 335                                 consecutive_spaces++;
 336                                 if (consecutive_spaces == 8) {
 337                                         *dst++ = '\t';
 338                                         consecutive_spaces = 0;
 339                                 }
 340                         }
 341                 }
 342                 while (0 < consecutive_spaces--)
 343                         *dst++ = ' ';
 344                 len -= last;
 345                 src += last;
 346                 fixed = 1;
 347         }
 348
 349         memcpy(dst, src, len);
 350         if (add_cr_to_tail)
 351                 dst[len++] = '\r';
 352         if (add_nl_to_tail)
 353                 dst[len++] = '\n';
 354         if (fixed && error_count)
 355                 (*error_count)++;
 356         return dst + len - buf;
 357 }